Product quality assessment and pricing metric system

Abstract

The systems and methods described herein can non-destructively identify a current produce quality. An evaluation system can receive image data of a produce at a first time and use machine learning techniques to identify the type of produce. The produce can be sent to an end-consumer retail environment. A quality assessment system can receive a request for a current quality of a particular produce from a user device in the retail environment at a second time, including scan data of the produce. The quality assessment system can identify the produce, retrieve from a data store a machine learning trained model for identifying quality characteristics of the produce, determine a current quality of the produce based on applying the model to the scan data, and transmit the current quality to the user device. The produce price can also be dynamically adjusted based on the current quality of the produce.

Description

[Technical field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to U.S. Provisional Patent Application No. 63 / 173,774, filed April 12, 2021, the disclosure of which is incorporated by reference in its entirety.

[0002] Technical Field This specification describes devices, systems, and methods related to identifying and using product metrics, such as product quality and pricing metrics. [Background technology]

[0003] background

[0003] Produce such as fruits and vegetables and / or other short-life goods (e.g., meat) may have different quality metrics that may affect suitability for consumption and value in the supply chain. Many different stakeholders throughout the supply chain have an interest in assessing the quality of such produce. The quality of the produce may affect the pricing and consumer purchasing decisions of such produce. For example, consumers may prefer to purchase ripe produce over out-of-season produce. Out-of-season produce may not be purchased or otherwise consumed by consumers, which may lead to produce-based waste. Furthermore, when produce has the same price regardless of quality, consumers may be inclined to purchase higher quality produce over lower quality food, which may also lead to waste of lower quality produce.

[0004]

[0004] Defining the quality of a produce objectively and quantitatively can be difficult. Relevant stakeholders in the supply chain can identify the quality of a produce by observing the visible characteristics of the produce. However, stakeholders may not be able to identify less visible characteristics that may affect the quality of the produce. Furthermore, destructive quality measurements can be made by touching, crushing, puncturing, slicing, and / or peeling a portion of the produce. Such destructive measurements may reduce or damage the quality of the produce. Furthermore, destructive measurements may not be accurate as to whether a consumer would want to purchase or otherwise consume the produce at this time. Summary of the Invention

[0005] overview

[0005] This specification generally describes techniques for identifying and using produce metrics, such as quality metrics (e.g., ripeness metrics, freshness metrics, taste metrics) and pricing metrics (e.g., price of the produce at its current quality level), that may change as the produce (e.g., fruits, vegetables) ripens or otherwise changes over time. For example, different types of produce may ripen at different rates, and the same type of produce may ripen at different rates depending on where the produce was grown, the manner in which the produce was transported (e.g., refrigerated container, non-refrigerated container), and / or other factors. As a result, it may be difficult to accurately assess produce quality without using techniques that may ruin or otherwise damage the produce, such as invasive measurement techniques (e.g., cutting open fruit to assess internal quality). The disclosed technology can generate and use data models that can be used to correlate non-invasive produce assessments (e.g., image-based produce assessments, analysis of volatiles and compounds emitted by the produce) with produce metrics, dynamically assess produce metrics as the produce ripens or otherwise changes over time, and further predict future produce metrics. Such non-invasive produce assessments can be performed, for example, at a single point in the supply chain of the produce item, and the produce item can have a unique identifier associated with the assessment (e.g., a sticker with a QR code on the produce item). The assessment can then be retrieved using the produce item's unique identifier and used in combination with the corresponding data model to dynamically identify produce metrics for the produce item at later points in the supply chain, as well as predict produce metrics for the produce item at future times without requiring further assessments. Such non-invasive produce assessment may be performed at a single point in the supply chain where volatiles emitted by the produce can be collected and analyzed to identify produce quality metrics and predict future produce metrics to determine supply chain changes (e.g., produce pricing, produce discounts, amount of time until ripe for eating, etc.) The volatiles can be modeled for shelf life and / or quality, which in turn can predict the shelf life and / or quality of the produce.Such produce assessments and metrics may be used in a variety of different applications, such as enabling consumers to more accurately assess produce quality in supermarkets or other retail environments, enabling retailers to dynamically adjust produce pricing based on quality, and / or other uses as described throughout this specification.

[0006]

[0006] As an illustrative example, a hyperspectral image of an avocado may be captured and used by a computing system to predict the number of days until the avocado is "ripe". The avocado may receive a sticker with a barcode that uniquely identifies it. The prediction and the unique identifier may be stored in a data store and used for subsequent quality and pricing predictions. As another illustrative example, the avocado may be placed in a produce bin for storage, movement along a supply chain, and / or point-of-sale, such as at a supermarket. The produce bin may be configured with one or more sensors that may detect / collect different types of volatiles and other compounds emitted by the avocado. The collected volatiles may be measured and / or analyzed by the one or more sensors, which may transmit the volatile measurement information to the computing system. Using one or more models trained with machine learning techniques to assess the volatiles and correlate the volatiles to quality metrics, the computing system may predict various quality metrics of the avocado, such as the number of days until the avocado is ripe, shelf life, taste, etc. In some implementations, a volatiles assessment may be performed once early in the supply chain of an avocado (or a group of avocados in the same produce bin), so that output from the assessment can be used to make upstream adjustments / changes to the supply chain of that avocado or group of avocados.

[0007]

[0007] For example, the price of that particular avocado (or a group of avocados in the same produce bin as that particular avocado) may be adjusted by the computing system based on the above assessments and predictions. The price may be adjusted to motivate consumers to purchase the avocado as it approaches its expiration date or becomes ripe. For example, an avocado may have a higher price when it is unripe or not yet ripe. If the avocado is predicted to become ripe or is overripe, the price of the avocado may be reduced. The reduced price motivates consumers to purchase the avocado, thereby reducing the likelihood that the avocado will go to waste.

[0008]

[0008] Produce, for example avocados, can be priced according to size. An average price point can be set, but some produce may be worth more than the average price and some less than the average price. This can lead to an adverse selection problem. Consumers may first select produce that are worth more than the average price, and produce that are worth less than the average price may not be purchased or may not be purchased until all produce above the average price has been purchased. This can lead to a significant waste of the produce base. On the other hand, the disclosed technology can provide real-time assessment of the quality of the produce and discounts on produce below the average price. When consumers see a discount, they may be inclined to purchase the discounted produce, thereby increasing the overall average quality of the remaining produce on the shelf. This can also increase the perceived quality of the produce section in the supermarket.

[0009]

[0009] Although the illustrative example is described with reference to avocados, the disclosed technology may be applied to a variety of produce types. For example, the disclosed technology may be used to non-invasively assess a variety of different quality metrics of apples, citrus fruits, berries, vegetables, and other types of produce sold at stores. Furthermore, different models may be trained to assess quality metrics associated with different types of produce, and the different trained models may be used to assess quality metrics associated with different types of produce.

[0010]

[0010] In addition to the embodiments in the accompanying claims and described above, the following numbered embodiments are also innovative.

[0011]

[0011] Embodiment 1 is a system for identifying a quality level of a food product using non-contact assessment of the food product, the system being a quality assessment system comprising: receiving food data, the food data including image data of the particular food product; identifying a current quality level of the particular food product based on providing the food data as input to a food quality model, the food quality model being trained using image data of other foods having distinguishable quality characteristics, the other foods being of the same food type as the particular food product; identifying a number of days until the particular food product is ready to eat based on the current quality level of the particular food product; and determining a number of days until the particular food product is ready to eat based on the measured volatiles of the particular food product. and determining at least one supply chain change for the particular food product based on a current quality level of the particular food product and the adjusted days until eating; and a display device in communication with the quality assessment system, the display device configured to: request information about the quality of the particular food product from the quality assessment system, the information including at least one of the current quality level of the particular food product, the days until eating of the particular food product, the adjusted days until eating, and the at least one supply chain change; and visually present the requested information on a display in an end consumer retail environment.

[0012]

[0012] Embodiment 2 is the system of embodiment 1, wherein the quality assessment system is further configured to generate a ripeness curve for the particular food based at least in part on the current quality level of the particular food and the measured volatiles, the ripeness curve indicating the current ripeness of the particular food and the predicted ripeness level of the particular food over a predetermined period of time.

[0013]

[0013] Embodiment 3 is the system of embodiment 1 or 2, wherein the quality assessment system is configured to generate a maturity curve for a batch of food including a particular food product, and the batch of food is of the same type as the particular food product.

[0014]

[0014] Embodiment 4 is a system of any one of embodiments 1 to 3, further including a container configured to hold a quantity of food, including a particular food product, the container including one or more sensors attached to one or more sides of the container, one or more of the sensors configured to detect volatile compounds released by the food in the container, a display screen attached to one or more sides of the container and configured to dynamically display information about the food in the container, and a communications interface configured to communicate with a quality assessment system over one or more networks.

[0015]

[0015] Embodiment 5 is a system of any one of embodiments 1 to 4, wherein the quality assessment system is configured to send an output indicating at least one supply chain change for a particular food product to a display screen of the container for presentation.

[0016]

[0016] Embodiment 6 is a system of any one of embodiments 1 to 5, wherein at least one supply chain change is a price adjustment of the food in the container, and the quality assessment system is configured to determine a price adjustment of the food in the container based at least in part on a current quality level of the particular food, and to transmit the adjusted price of the food in the container to a display screen of the container for presentation.

[0017]

[0017] Embodiment 7 is a system of any one of embodiments 1 to 6, wherein the quality assessment system is configured to receive data from one or more sensors indicative of volatile compounds detected by the one or more sensors, provide the data indicative of the volatile compounds as input to a food quality model, the food quality model being trained to correlate volatile compounds emitted by other foods with quality characteristics of the other foods, the quality characteristics being annotated in the training data of the other foods, and receive as output from the food quality model a current quality level of the particular food based on the data indicative of the volatile compounds.

[0018]

[0018] Embodiment 8 is a system of any one of embodiments 1 to 7, wherein determining at least one supply chain change includes determining at least one sorting arrangement of the container relative to (i) the food in the container and (ii) other containers containing other types of food in the end consumer retail environment.

[0019]

[0019] Embodiment 9 is a system of any one of embodiments 1 to 8, wherein the sorting arrangement includes placing ripe foods closer to the front of the container and less ripe foods closer to the rear of the container.

[0020]

[0020] Embodiment 10 is a system of any one of embodiments 1 to 9, wherein the sorting arrangement includes placing containers having more ripe food closer to an entrance to the end consumer retail environment and placing containers having more unripe food further away from the entrance to the end consumer retail environment.

[0021]

[0021] Embodiment 11 is a system of any one of embodiments 1 to 10, wherein determining at least one supply chain change includes adjusting a price of food in the container based on a current quality level of the particular food in the container.

[0022]

[0022] Embodiment 12 is a system of any one of embodiments 1 to 11, wherein the quality assessment system is configured to (i) increase the price of the food in the container based on a determination that the current quality level is a ripeness level below the threshold ripeness value, or (ii) decrease the price of the food in the container based on a determination that the current quality level is a ripeness level above the threshold ripeness value.

[0023]

[0023] Embodiment 13 is a system of any one of embodiments 1 to 12, wherein determining one or more supply chain changes includes determining a shelf life of food in a container based on a current quality level.

[0024]

[0024] Example 14 is the system of any one of Examples 1 to 13, wherein the food quality model is a model that predicts ripeness of the food in the container based on the detected volatile compounds.

[0025]

[0025] Example 15 is the system of any one of Examples 1 to 14, wherein the food quality model is a model that predicts the firmness of the food in the container based on the detected volatile compounds.

[0026]

[0026] Embodiment 16 is the system of any one of embodiments 1 to 15, wherein the food quality model is a model that predicts the taste of the food in the container based on the detected volatile compounds.

[0027]

[0027] Embodiment 17 is the system of any one of embodiments 1 to 16, wherein the food quality model is a model that predicts the juiciness of the food in the container based on the detected volatile compounds.

[0028]

[0028] An eighteenth embodiment is the system of any one of the first to seventeenth embodiments, in which the display screen is a display device.

[0029]

[0029] Example 19 is the system of any one of Examples 1-18, wherein the one or more sides of the container include a front wall, a back wall, a first side wall, a second side wall, and a bottom of the container.

[0030]

[0030] Embodiment 20 is a system of any one of embodiments 1 to 19, wherein the container further includes a local power source that is part of the container and is configured to provide power to the one or more sensors and the display screen.

[0031]

[0031] Embodiment 21 is a system of any one of embodiments 1 to 20, further including a rating calculation system, wherein the rating calculation system is configured to: receive image data of the particular food from the imaging device; identify a type of the particular food based on applying a food identification model to the image data; determine an initial quality level of the particular food based on the identified type of the particular food; generate a product identifier that uniquely identifies the particular food, wherein information associated with the product identifier corresponds to the initial quality level of the particular food; generate instructions to transport the particular food to an end consumer retail environment based at least in part on the initial quality level of the particular food; and return to at least one party in the supply chain associated with the particular food (i) instructions for execution by at least one party in the supply chain, and (ii) the product identifier.

[0032]

[0032] Embodiment 22 is a system of any one of embodiments 1 to 21, wherein the valuation calculation system is configured to determine an initial price for a particular food product based at least in part on an initial quality level of the particular food product, the type of the particular food product, and preferences of the end consumer retail environment.

[0033]

[0033] Embodiment 23 is the system of any one of embodiments 1 to 22, wherein the specific food is avocado, citrus fruit, berry, orange, apple, mango, or vegetable.

[0034]

[0034] Embodiment 24 is a system of any one of embodiments 1 to 23, wherein the quality assessment system is configured to receive a request for a current quality level of a particular food product from a user device in an end consumer retail environment, the request including scan data of the particular food product, the scan data including image data of the particular food product and at least one unique identifier attached to the particular food product, and to transmit to the user device the current quality level of the particular food product for presentation in a mobile application on the user device.

[0035]

[0035] Example 25 is the system of any one of Examples 1 to 24, wherein the display device is a consumer user device in an end-consumer retail environment.

[0036]

[0036] Example 26 is the system of any one of Examples 1-25, wherein the display device is a visual display on a container configured to contain a particular food product in an end-consumer retail environment.

[0037]

[0037] Example 27 is the system of any one of Examples 1-26, wherein the display device is a visual display located near the particular food item within the end consumer retail environment.

[0038]

[0038] Embodiment 28 is a system of any one of embodiments 1 to 27, further including at least one volatile substance sensor, the at least one volatile substance sensor configured to detect volatile substances emitted by a particular food product in an end consumer retail environment and transmit data indicative of the detected volatile substances to a quality assessment system.

[0039]

[0039] Embodiment 29 is a system of any one of embodiments 1 to 28, wherein the quality assessment system is configured to receive data indicative of detected volatiles from at least one volatile sensor, and process the received data to generate measured volatiles of a particular food.

[0040]

[0040] Embodiment 30 is a container configured to preserve food quality, the container including one or more sensors mounted on one or more sides of the container, one or more of the sensors configured to detect volatile compounds released by the food in the container, a display screen mounted on one or more sides of the container and configured to dynamically display information about the food in the container, a communications interface configured to communicate with a quality assessment system over one or more networks, and a controller in communication with the one or more sensors, the display screen, and the communications interface, the controller configured to update the information displayed on the display screen based on quality metrics of the food in the container.

[0041]

[0041] Example 31 is the container of example 30, wherein the container further includes a local power source configured to provide power to the one or more sensors, the display screen, the controller, and the communication interface.

[0042]

[0042] Embodiment 32 is a container of embodiment 30 or 31, further comprising at least one sensor positioned a threshold distance away from the container and configured to detect volatile compounds emitted by food in the container.

[0043]

[0043] Embodiment 33 is the container of any one of embodiments 30-32, wherein the threshold distance is within the range of 1 to 6 inches.

[0044]

[0044] Embodiment 34 is a container of any one of embodiments 30 to 33, wherein the quality metrics of the food in the container are determined by a computer system in communication with the container, and the computer system is configured to perform operations including receiving data from the controller indicative of volatile compounds detected by one or more sensors; providing the data indicative of the volatile compounds as input to a food quality model, the food quality model being trained to correlate volatile compounds emitted by other foods of the same food type as the food in the container with quality characteristics of the other foods; receiving the quality metrics of the food in the container as output from the food quality model; and transmitting, via the communications interface, an output to the controller for presentation on a display screen, the output being indicative of the quality metrics of the food.

[0045]

[0045] Embodiment 35 is the container of any one of embodiments 30-34, wherein the output indicative of a quality metric of the food product includes a current ripeness level of the food product.

[0046]

[0046] Embodiment 36 is the container of any one of embodiments 30 to 35, wherein the output indicative of a quality metric of the food product includes a current firmness of the food product.

[0047]

[0047] Embodiment 37 is the container of any one of embodiments 30 to 36, wherein the output indicative of a quality metric of the food product includes a current readiness of the food product.

[0048]

[0048] Embodiment 38 is the container of any one of embodiments 30 to 37, wherein the output indicative of a quality metric of the food product includes a current taste of the food product.

[0049]

[0049] Embodiment 39 is the container of any one of embodiments 30-38, wherein the output indicative of a quality metric of the food product includes a current sweetness of the food product.

[0050]

[0050] Embodiment 40 is a container of any one of embodiments 30 to 39, wherein the operations further include determining at least one supply chain change based on quality metrics of the food in the container, and sending the at least one supply chain change to the controller via the communication interface for execution.

[0051]

[0051] Embodiment 41 is a container of any one of embodiments 30 to 40, wherein at least one supply chain change is adjusting a price of the food in the container based on a quality metric, and the controller is configured to (i) generate an updated output including the adjusted price of the food in the container, and (ii) instruct the display screen to present the updated output.

[0052]

[0052] Embodiment 42 is a container of any one of embodiments 30 to 41, wherein the food quality model is trained to model data indicative of volatile compounds to a quality curve of the food in the container.

[0053]

[0053] Embodiment 43 is a container of any one of embodiments 30 to 42, wherein the food quality model has been trained using a process including receiving, by a computer system, training data of other foods, the training data including (i) food observations, (ii) food history information, (iii) destruction measurements of the other foods, and (iv) detected volatile compounds from the other foods, where at least one of (i) to (iv) is labeled and annotated with food features; applying machine learning techniques to the training data to generate a food quality model based on mapping the detected volatile compounds from the other foods to the food features; and outputting the food quality model for use at run time.

[0054]

[0054] Embodiment 55 is a case configured to preserve food quality, the case including a physical structure having one or more levels for storing food groups; one or more sensors attached to the case, one or more of the sensors configured to detect volatile compounds emitted by the food groups at one or more levels of the physical structure; one or more display devices configured to dynamically display information about the food groups in the case; a hose structure positioned around the food groups; a vacuum system configured to draw air from around the food groups through the hose structure and direct a concentration of volatile compounds emitted by the food groups towards at least one of the one or more sensors; and a controller in communication with the one or more sensors, the one or more display devices and the vacuum system, the controller configured to update information displayed on the one or more display devices based on quality metrics of the food groups in the case, the quality metrics of the food groups being determined at least in part based on the volatile compounds emitted by the food groups.

[0055]

[0055] Embodiment 45 is a case of embodiment 44, in which the physical structure is a rack.

[0056]

[0056] Embodiment 46 is a case of embodiment 44 or 45, where one or more of the levels are shelves.

[0057]

[0057] Embodiment 47 is the case of any one of embodiments 44 to 46, wherein the case is a refrigeration unit.

[0058]

[0058] An embodiment 48 is the case of any one of the embodiments 44 to 47, in which the case is a cabinet.

[0059]

[0059] Embodiment 49 is the case of any one of embodiments 44 to 48, wherein the case is a food aging cabinet.

[0060]

[0060] Embodiment 50 is the case of any one of embodiments 44-49, wherein the case is a food aging enclosure.

[0061]

[0061] Embodiment 51 is the case of any one of embodiments 44 to 50, in which the one or more sensor groups are attached to one or more sides of the case proximate to the food group.

[0062]

[0062] Embodiment 52 is any one of embodiments 44 to 51, in which the one or more sensor groups are attached to a physical structure in proximity to the food group.

[0063]

[0063] Embodiment 53 is any one of embodiments 44 to 52, in which the one or more sensor groups are mounted on at least one of the one or more levels proximate the food group.

[0064]

[0064] Embodiment 54 is any one of embodiments 44 to 53, further including one or more containers located at one or more levels, each of the one or more containers configured to contain a respective one of the food groups.

[0065]

[0065] Embodiment 55 is a system for analyzing food quality using non-contact assessment of food products, the system including a case configured to hold food groups, a volatile substance assessment system configured to collect volatile compounds emitted by the food groups within the case and process the collected volatile compounds to generate volatile substance data for the food groups, a tubing structure positioned around the case and in fluid communication with the volatile substance assessment system, the tubing structure configured to transmit the volatile compounds emitted by the food groups within the case to the volatile substance assessment system, a controller configured to control operation of the volatile substance assessment system, and a communications interface configured to communicate with the quality assessment system over one or more networks, wherein the quality assessment system is configured to identify one or more quality metrics for the food groups based at least in part on the volatile substance data.

[0066]

[0066] Embodiment 56 is a system of embodiment 55, further including a vacuum system in communication with the volatile substance assessment system, the vacuum system configured to draw volatile compounds emitted by the food groups in the case through the tube structure and toward the volatile substance assessment system for collection.

[0067]

[0067] Embodiment 57 is the system of embodiment 55 or 56, wherein the vacuum system includes a pump configured to draw the volatile compounds through the tubing and towards the volatile material assessment system for collection.

[0068]

[0068] Embodiment 58 is the system of any one of embodiments 55 to 57, wherein the vacuum system includes a fan configured to draw the volatile compounds through the tube structure and towards the volatile substance assessment system for collection.

[0069]

[0069] Embodiment 59 is a system of any one of embodiments 55 to 58, further comprising a valve in fluid communication with the tube structure and the volatile substance assessment system, and the controller is configured to open the valve to allow volatile compounds to pass from the tube structure into the volatile substance assessment system for collection.

[0070]

[0070] Embodiment 60 is a system of any one of embodiments 55 to 59, wherein the controller is configured to close the valve based on a determination that a predetermined threshold amount of volatile compounds has been collected by the volatile substance assessment system.

[0071]

[0071] Embodiment 61 is a system of any one of embodiments 55 to 60, wherein the controller is configured to generate instructions that, when executed, cause the volatile substance assessment system to process the collected volatile compounds and generate volatile substance data.

[0072]

[0072] Embodiment 62 is the system of any one of embodiments 55 to 61, wherein the valve is a switch.

[0073]

[0073] Embodiment 63 is a system of any one of embodiments 55 to 62, further including a display device in communication with the volatile substance assessment system over one or more networks, the display device configured to present information about the food group on a display screen.

[0074]

[0074] Embodiment 64 is the system of any one of embodiments 55 to 63, in which the display device is located next to the case.

[0075]

[0075] Embodiment 65 is the system of any one of embodiments 55 to 64, in which the display device is attached to the case.

[0076]

[0076] Embodiment 66 is a system of any one of embodiments 55 to 65, wherein the display device communicates with the quality assessment system over one or more networks, and the display device is configured to present one or more quality metrics of the food groups in the case.

[0077]

[0077] Example 67 is the system of any one of Examples 55 to 66, wherein the case includes one or more openings in fluid communication with one or more tubes of the tube structure.

[0078]

[0078] Embodiment 68 is the system of any one of embodiments 55 to 67, wherein the one or more openings are integrated into a bottom surface of the case.

[0079]

[0079] Embodiment 69 is the system of any one of embodiments 55-68, wherein the one or more tubes are sized in the range of approximately 1 / 16 inch to 1 / 4 inch.

[0080]

[0080] Embodiment 70 is a system of any one of embodiments 55 to 69, further comprising a vacuum system configured to draw air through one or more tubes of the tube structure for collection and to transmit the volatile compounds emitted by the food groups through one or more openings in the case, through the one or more tubes and to a volatile substance assessment system.

[0081]

[0081] Embodiment 71 is the system of any one of embodiments 55 to 70, wherein the volatile substance assessment system is an electronic nose.

[0082]

[0082] Embodiment 72 is the system of any one of embodiments 55 to 71, wherein the volatile substance assessment system includes one or more sensors for detecting volatile compounds emitted by the food group.

[0083]

[0083] Embodiment 73 is a system of any one of embodiments 55 to 72, wherein each of the one or more sensors is configured to detect a different type of volatile compound among the volatile compounds emitted by the food group.

[0084]

[0084] Embodiment 74 is the system of any one of embodiments 55 to 73, wherein the volatile substance assessment system is a gas chromatography machine.

[0085]

[0085] Embodiment 75 is an apparatus for analyzing food quality using non-contact assessment of food products, the apparatus including a case configured to hold a group of food products, the case including one or more sensors configured to collect volatile compounds emitted by the food groups within the case, a vacuum system configured to draw the volatile compounds emitted by the food groups within the case towards the one or more sensors, and a controller configured to control operation of the vacuum system.

[0086]

[0086] Embodiment 76 is the apparatus of embodiment 75, wherein the vacuum system includes a pump.

[0087]

[0087] Embodiment 77 is the apparatus of embodiment 75 or 76, wherein the vacuum system includes a fan.

[0088]

[0088] Embodiment 78 is an apparatus of any one of embodiments 75 to 77, wherein the one or more sensors are selected from a group of sensors, the group of sensors including at least one of a nanocomposite sensor, an electromechanical sensor, a metal oxide semiconductor, a photoionization sensor, and a non-dispersive infrared sensor.

[0089]

[0089] Embodiment 79 is an apparatus of any one of embodiments 75 to 78, further including a display device in communication with the controller over one or more networks, the controller being configured to generate instructions that, when executed, cause the display device to present information about the food groups in the case, the information being based, at least in part, on volatile compounds collected by the one or more sensors.

[0090]

[0090] Embodiment 80 is an apparatus of any one of embodiments 75 to 79, wherein the controller is configured to generate volatile substance data for the food groups in the case based on volatile compounds collected by one or more sensors.

[0091]

[0091] Embodiment 81 is an apparatus of any one of embodiments 75 to 80, wherein the controller is configured to transmit the volatile substance data to a quality assessment system configured to identify one or more quality metrics for the food group based at least in part on the volatile substance data.

[0092]

[0092] Embodiment 82 is the device of any one of embodiments 75 to 81, wherein the controller is configured to identify one or more quality metrics for the food group based at least in part on the volatile substance data.

[0093]

[0093] Embodiment 83 is a method for dynamically identifying a quality level of a food product using non-contact assessment of the food product, the method including: receiving, by a computing system, food data, where the food data includes image data of the particular food product; determining, by the computing system, a current quality level of the particular food product based on providing the food data as input to a food quality model, where the food quality model is trained using image data of other foods having distinguishable quality characteristics, where the other foods are of the same food type as the particular food product; and determining, by the computing system and based on the current quality level of the particular food product, a number of days until the particular food product is ready to eat. adjusting, by the computing system, the identified days until eating based on the measured volatiles of the particular food product; determining, by the computing system, at least one supply chain change for the particular food product based on the current quality level and the adjusted days until eating of the particular food product; and transmitting, by the computing system, to a display device, information about the quality of the particular food product to be visually presented on a display in the end consumer retail environment, the information including at least one of the current quality level of the particular food product, the days until eating of the particular food product, the adjusted days until eating, and the at least one supply chain change.

[0094]

[0094] Embodiment 84 is the method according to any one of embodiments 1 to 82.

[0095]

[0095] Certain embodiments described herein may include a system for identifying a quality level of a food product using non-contact assessment of the food product, the system including an assessment calculation system and a quality assessment system. The assessment calculation system may receive image data of the food product from an imaging device at a first time and identify a type of the food product based on applying a first set of trained models to the image data. Each trained model may be trained using at least (i) image data of other foods having a first set of food type identification features, and (ii) image data of other foods having a second set of food type features. The assessment system may also generate a product identifier that uniquely identifies the food product, generate instructions to transport the food product to an end-consumer retail environment, and transmit the product identifier of the food product to the quality assessment system. The quality assessment system may store the product identifier in a data store at a second time and receive a request for a current quality level of a particular food product from a user device in the end-consumer retail environment. The request may include scan data of the particular food product. The quality assessment system can identify a type of the particular food based on applying the first set of trained models to the scan data and can retrieve a second set of trained models from the data store based on identifying the particular food as being of the same type as the particular food. Each trained model can be trained using (i) image data of other foods having poor quality characteristics and (ii) image data of other foods having good quality characteristics, the other foods being of the same food type as the particular food. The quality assessment system can identify a current quality level of the particular food based on applying the second set of trained models to the scan data and can transmit the current quality level of the particular food to the user device.

[0096] In some implementations, the embodiments described herein may include one or more of the following features. For example, the first time may be earlier than the second time. The scan data may include image data of the particular food item. The imaging device may be a hyperspectral imaging device. The product identifier may be a barcode, SKU, QR code, sticker, stamp, or label.

[0097]

[0097] As another example, identifying the type of the particular food product may include detecting a product identifier in the scan data using object recognition techniques, decoding the product identifier, and matching the decoded product identifier of the particular food product with the product identifiers of the food products stored in the data store. Further, the first and second sets of trained models may be trained using a convolutional neural network (CNN). At least one of the image data and the scan data may include color data, an RGB color model, and hyperspectral data.

[0098]

[0098] As yet another example, the quality assessment system may determine that a particular food product is suitable for consumption based on a current quality level of the particular food product exceeding a threshold value. The assessment calculation system may also determine an initial quality level of the food product based on applying a second set of trained models to the image data based on the identified type of food product. Determining the initial quality level of the food product may include predicting the number of days until the food product is ripe for consumption.

[0099]

[0099] In some implementations, the valuation calculation system can also identify the type of food product based on supply chain information at the food product's current location. The supply chain information can include production location, transit conditions, and storage conditions. The valuation calculation system can also determine an initial price for the food product. The initial price can be based on market conditions, the type of food product, and end-consumer retail environment preferences.

[0100]

[0100] As another example, determining a current quality of the particular food product may include generating a quality curve. The quality curve may indicate a maturity stage of the particular food product over a predetermined time period. The predetermined time period may include a first time, a second time, and an additional time.

[0101]

[0101] As yet another example, the quality assessment system may detect an environment of the particular food in the scan data using object recognition techniques, predict a shelf life of the particular food based on applying the second set of trained models to the scan data, compare a location of the particular food in the detected environment to the predicted shelf life of the particular food, and determine whether the location of the particular food in the detected environment corresponds to the predicted shelf life of the particular food. The quality assessment system may generate an output of the predicted shelf life of the particular food based on determining that the predicted shelf life corresponds to a current location of the particular food in the detected environment, and improve one or more models in the second set of trained models to predict the shelf life of the particular food based on training the one or more models using the scan data.

[0102]

[0102] In some implementations, the quality assessment system may also receive, from a user device, image data of a store shelf having particular food products in an end consumer retail environment, identify a product identifier for each of the plurality of food products, determine a current quality level of the food product based on the product identifier for each of the plurality of food products, and generate an augmented reality view of the image data for display on the user device in which one or more of the plurality of food products are highlighted based on the current quality level of each of the plurality of food products. Each of the plurality of food products having the highest current quality level may be highlighted in the augmented reality view of the image data. Each of the plurality of food products having the lowest current quality level may also be highlighted in the augmented reality view of the image data.

[0103]

[0103] As another example, the quality assessment system may receive an indication from the user device at a third time that a user of the user device has purchased a particular food item, identify a quality level of the purchased food item at a third time after the second time, and predict when the purchased food item will be ready for consumption by the user based on the quality level of the purchased food item at the third time. If the purchased food item is no longer ready for consumption, the quality assessment system may identify one or more recipes that include the purchased food item and generate an output indicating one or more recipes that can be used (i) when the purchased food item is ready for consumption and (ii) when the purchased food item is no longer ready for consumption, for display on the user device.

[0104]

[0104] The preferred embodiment herein may also include a system for dynamically adjusting the price of a food product using non-contact assessment of the food product. The system may include a quality assessment system and a pricing calculation system. The quality assessment system may receive a request for a current quality level of a particular food product from a user device in an end consumer retail environment. The request may include scan data of the particular food product. The quality assessment system may identify a type of the particular food product based on applying a first set of trained models to the scan data. Each trained model may be trained using at least (i) image data of other food products having a first set of food type identifying features, and (ii) image data of other food products having a second set of food type features. The quality assessment system may then retrieve from the data store, based on the identification of the type of the particular food product, past quality information for foods of the same type as the particular food product and a second set of trained models trained to identify quality features of foods of the same type as the particular food product. Each of the trained models may be trained using (i) image data of foods having poor quality features, and (ii) image data of foods having good quality features. The foods may be of the same food type as the particular food product. The quality assessment system may identify a current quality level of the particular food product based on applying the historical quality information and the second set of trained models to the scan data, transmit the current quality level of the particular food product to the user device for display on the user device, and transmit the current quality level of the particular food product to the pricing calculation system. The pricing calculation system may receive the current quality level of the particular food product from the quality assessment system, and retrieve from the pricing model data store a third set of trained models that may be trained to determine a price adjustment for the particular food product based on the current quality level of the particular food product. Each trained model may be trained using at least (i) pricing data of other foods having less favorable quality levels, and (ii) image data of other foods having higher quality levels.The pricing calculation system can determine a price adjustment for the particular food product based on applying the third set of trained models to a current quality level of the particular food product and transmit the price adjustment to the user device for display on the user device.

[0105]

[0105] In some implementations, the embodiments described herein may include one or more of the following features. For example, determining a price adjustment for the particular food product may include decreasing the price of the particular food product if the current quality level of the particular food product is not within a threshold range. Determining a price adjustment for the particular food product may also include increasing the price of the particular food product if the current quality level of the particular food product is within a threshold range. The current quality level of the particular food product may not be within a threshold range if the particular food product is past its prime. The current quality of the particular food product may be within a threshold range if the particular food product is approaching prime, ripe, or ripe for eating.

[0106]

[0106] Furthermore, the specific food may be an avocado, a fruit, a vegetable, or a meat. The scan data may include image data of the specific food. The scan data may also include a product identifier of the specific food. The quality assessment system may detect the product identifier in the scan data using object recognition techniques and identify the specific food based on matching the decoded product identifier of the specific food with product identifiers of other foods of the same type as the specific food stored in the data store.

[0107]

[0107] In some implementations, determining a price adjustment for the particular food product based on applying the third set of trained models to a current quality level of the particular food product may include predicting a remaining marketable shelf life of the particular food product, determining whether the remaining marketable shelf life of the particular food product exceeds a threshold range, increasing the price of the particular food product based on a determination that the remaining marketable shelf life of the particular food product exceeds the threshold range, and decreasing the price of the particular food product based on a determination that the remaining marketable shelf life of the particular food product does not exceed the threshold range.

[0108]

[0108] The system may also include a store point of sale (POS) terminal in communication with the quality assessment system at the checkout lane. The POS terminal may include receiving a current quality level of the particular food item from the query assessment system. The particular food item may be scanned at the POS terminal and purchased by the consumer. The POS terminal may retrieve pricing information corresponding to food items of the same type as the particular food item from a data store, determine a price adjustment for the particular food item based on the pricing information and the current quality of the particular food item, and implement the price adjustment for the particular food item. The price adjustment may be automatically applied to the consumer's bill.

[0109]

[0109] As another example, sending a price adjustment to a user device may include generating a coupon that can be applied during checkout to purchase a particular food item.

[0110]

[0110] The preferred embodiments described herein may also include a system for classifying food products based on their current quality level using non-contact assessment of the food products. The system may include a quality assessment system and a user device. The quality assessment system may receive a request for a current quality level of a particular food product from a user device in an end consumer retail environment. The request may include scan data of a plurality of food products. The quality assessment system may identify a type of food product based on applying a first set of trained models with each of the plurality of food products to the scan data. The first set of trained models may be trained to identify and decode product identifiers of a plurality of other food products. The quality assessment system may retrieve from the data store, for each of the plurality of food products, a type of food product, past quality information for foods of the same type as the food product, and a second set of trained models that may be trained to identify quality characteristics of foods of the same type as the food product. Each trained model may be trained using (i) image data of foods having poor quality characteristics, and (ii) image data of foods having good quality characteristics. The foods may be of the same food type as the particular food product. The quality assessment system can identify a current quality level of the food product based on applying the second set of trained models to the scan data for each of the plurality of food products and transmit the current quality level of each of the plurality of food products to the user device for display on the user device. The user device can output the current quality level of each of the plurality of food products, rank the plurality of food products from highest current quality level to lowest current quality level, and determine a sorting order for the ranked plurality of food products. The sorting order can indicate that the food products having the highest current quality level may be located at the back of the store shelf and the food products having the lowest current quality level may be located at the front of the store shelf.

[0111]

[0111] In some implementations, the embodiments described herein may include one or more of the following features. For example, the highest current quality level may indicate that the food product is at least one of: (i) not yet ripe, and (ii) having a maximum number of days remaining until it is ripe to eat. The lowest current quality level may indicate that the food product is at least one of: (i) past its ripeness, and (ii) having a minimum number of days remaining before it is no longer consumable.

[0112]

[0112] The disclosed technology may provide one or more of the following advantages. For example, the disclosed technology may provide for the elimination of product-based waste while maximizing value to relevant stakeholders in the supply chain. The quality of a product can be dynamically assessed in real time and used to determine price adjustments for that product. Products that may be deteriorating in quality can be discounted, providing an incentive to consumers to purchase them over products that are improving in quality. When products that are deteriorating in quality are purchased, product-based waste can be reduced or eliminated.

[0113]

[0113] The stakeholder furthest up the supply chain who last touches the product can provide the greatest cost advantage to the end consumer. There can also be a market-equilibrium price for any given price of the product that is a function of the product's quality. This can be advantageous for the stakeholder furthest up the supply chain, since every time the product is handled, it adds a cost. For example, a producer who ships directly to a grocer can deliver at a lower cost than a producer who must handle the product and then deliver it to a marketer who delivers it to the grocer. The marketer can be an aggregator, justifying the cost by matching supply and demand. Because products are natural products and therefore variable in quality, it can be advantageous to dynamically adjust the price of individual products as described herein. Therefore, it can be desirable to price products according to their quality in order to minimize the cost of buying and selling (e.g., minimize labor costs).

[0114]

[0114] Furthermore, identifying and predicting shelf life and other quality metrics of produce close to the point of purchase can reduce the need to dynamically perform shelf life or other quality predictions throughout the supply chain. Predictions can be dynamically output, for example, to displays located near the produce to provide consumers with the information they need to make informed purchasing decisions, and / or can be dynamically output to a user computing device using a unique identifier for the produce (e.g., scanning a QR code associated with the container to facilitate retrieval of the predictive information). Predictions can also be used to make precise price adjustments at or near the point of purchase. Dynamically outputting predictions to in-store displays and / or to consumer user devices can drive purchasing behavior. The real-time analytics described herein also provide for dynamic pricing, rather than making price adjustments based on modeling produce measurements made earlier in the supply chain.

[0115]

[0115] As another example, a machine learning trained model and a robust set of stored data about the product can be used to more accurately and dynamically identify product quality in real time. The human eye can be prone to error when trying to observe subtle changes in the appearance of the product, which can negatively affect a person's ability to accurately assess product quality. Human observation can also be a tedious and time-consuming process. The disclosed technology provides for automatic accurate identification of product quality using a machine learning trained model and a robust set of stored data about the product. The disclosed technology can provide deeper analysis, thereby increasing efficiency and reducing human error that can result from visual observation and inspection of the product.

[0116]

[0116] As another example, the disclosed technology may be used to make appropriate supply chain changes to reduce produce-based waste. As mentioned above, prices may be adjusted based on the current assessment of produce quality. Furthermore, the placement of produce on the shelves may also be adjusted based on the current assessment of produce quality. A supermarket employee may, for example, request a quality assessment of produce that needs to be stocked on the shelves from a computing system. The computing system may return a quality value for each produce. The employee may then stock the shelves based on the quality values ​​of the produce. As an example, the employee may stock the front of the shelves with produce that has the lowest quality (e.g., overripe, past its prime, has some blemishes, etc.). The employee may place produce with the highest quality (e.g., still not ripe, approaching prime, without blemishes, etc.) towards the back of the shelves. As a result, consumers may select produce at the front of the shelves first, which may be advantageous to sell lower quality produce and avoid produce-based waste.

[0117]

[0117] Similarly, the disclosed technology provides accurate prediction of shelf life of produce to optimize distribution of produce to suppliers, retailers, distributors and / or consumers throughout the supply chain. Non-destructive measurements of produce can be captured during in-store purchase and / or anywhere along the supply chain, such as at a packing or storage facility, and used in combination with historical information about the produce and machine learning trained models to accurately determine the classification sequence of the produce, predict shelf life based on the assumed supply chain of the produce, and update the prediction based on deviations from the prediction. The disclosed technology can also be used to identify sensory characteristics of produce (especially aroma, taste, appearance) that guide consumer purchasing decisions. For example, the better tasting a produce, the more likely a consumer is to purchase that produce again. These characteristics may be difficult to detect or identify by simple human observation when the consumer is in the store. Thus, the disclosed technology provides predictions of flavor measurements of produce, helping consumers make more informed purchasing decisions.

[0118]

[0118] As yet another example, the disclosed technology can generate more robust quality and pricing assessments of products. Different models can be generated and trained using machine learning techniques and high-quality labeled training datasets to identify and score different quality features associated with different products. Thus, models can be trained to identify features that may be difficult for humans to observe and associate such features with product quality and pricing. Such robust quality and pricing assessments can be advantageous to more accurately and dynamically monitor product quality, adjust product pricing, and modify the supply chain.

[0119]

[0119] As another example, a retailer can see information about what produce has been scanned and associated discounts and / or pricing adjustments. Using such information, the retailer can determine (1) how efficiently the store is being marketed, (2) the quality of produce consumers are receiving, and (3) the consumer's ability to understand the price elasticity of quality. Such information can be used by the retailer to optimize its business.

[0120]

[0120] Furthermore, the disclosed technology provides improved accuracy of product quality and pricing assessment using the geometric learning trained model and non-invasive analysis of volatiles emitted by the product. Volatiles and other compounds emitted by the product can be captured and analyzed using the disclosed technology to determine shelf life and other quality metrics of the product. Such identification can then be used to predict the future quality of the product and make appropriate discounts and / or pricing adjustments upstream in the supply chain. Identification can be done at the beginning of the supply chain before the product is placed on the shelf for sale in a supermarket. Thus, volatile assessment can improve accuracy in predicting product quality early in the supply chain and provide consumers with accurate upstream information about the products they can purchase at the supermarket.

[0121]

[0121] As another example, produce shelves and / or containers can be embedded with sensors that collect information about volatiles and other compounds emitted by produce placed therein. Different sensors can be used to detect different volatiles emitted by produce, and different sensors can be configured to detect different volatiles. Furthermore, arrays of sensors can be used, which may have one or more sensors for each volatile or compound of interest. Such arrays can beneficially provide accurate volatile / compound collection. Such produce shelves and / or containers can be deployed in supermarkets at a low cost to the supermarket. These produce shelves and / or containers can provide accurate, real-time collection of volatiles and compounds emitted by produce, which can be assessed using the disclosed techniques to identify the current quality of the produce and the future quality of the produce. Furthermore, supermarkets can be manufactured with sensors and populated with produce shelves and / or containers that perform the techniques described throughout this disclosure.

[0122]

[0122] The disclosed technology also provides for generating a robust set of models trained using machine learning techniques to assess different quality metrics of different types of products. Each model can be trained using a robust training dataset to accurately identify current quality metrics and predict future quality metrics of a particular product type. During real-time assessment of a product, one or more models can be selected and applied based on the type of product to accurately and non-invasively assess the product and determine upstream supply chain changes for the product.

[0123]

[0123] The details of one or more implementations are set forth in the accompanying drawings and the description below. Other features and advantages will be apparent from the description and drawings, and from the claims. [Brief description of the drawings]

[0124] BRIEF DESCRIPTION OF THE DRAWINGS [Figure 1A]

[0124] Conceptual diagram of dynamic specification of product quality. [Figure 1B]

[0125] 1 is a conceptual diagram of a produce display system for identifying and presenting produce quality information to a user in a store. [Figure 1C]

[0126] FIG. 1 is a perspective view of a product container used to practice the techniques described herein. [Figure 1D]

[0127] FIG. 1D is a side view of the produce container of FIG. [Figure 1E]

[0128] FIG. 1D is a front view of the produce container of FIG. [Figure 1F]

[0129] FIG. 1D is a top view of the product container of FIG. [Diagram 2]

[0130] FIG. 1 is a conceptual diagram of dynamic determination of product pricing. [Diagram 3]

[0131] FIG. 1 is a conceptual diagram of dynamic adjustment of product pricing based on product quality. [Figure 4]

[0132] FIG. 1 is a conceptual diagram of product quality identification for consumers such as shoppers. [Figure 5A]

[0133] A conceptual diagram of store employees arranging shelves based on product quality. [Figure 5B]

[0134] 5B shows the display shelf of FIG. 5A arranged by a store employee. [Figure 5C]

[0135] FIG. 1 is a conceptual diagram of sorting of produce containers based on a quality assessment of the product within the container. [Figure 5D]

[0136] FIG. 1 is a conceptual diagram of product pricing decisions based on quality assessment of the product in each bin. [Figure 6]

[0137] FIG. 1 is a conceptual diagram of product identification and labeling. [Figure 7]

[0138] 1 is a conceptual diagram of identifying the current quality of a product. [Figure 8]

[0139] 1 is a graphical depiction of the same type of product with different quality levels at the current time. [Figure 9A]

[0140] FIG. 1 is a system diagram of components used to perform the techniques described throughout this disclosure. [Figure 9B]

[0140] A system diagram of components used to implement the techniques described throughout this disclosure. [Figure 10A]

[0141] 1 is a schematic flow chart of a process for evaluating a product. [Figure 10B]

[0142] 1 is a flow chart of a process for evaluating a produce based on an assessment of volatile substances emitted by the produce. [Figure 11]

[0143] 1 is a flowchart of a process for determining a price of a product. [Figure 12]

[0144] 1 is a flowchart of a process for improving a model used to determine product quality. [Figure 13]

[0145] 1 illustrates example components of a user device application interface. [Figure 14A]

[0146] 13 is a flowchart of a process for displaying products based on quality using augmented reality (AR) in a user device application. [Figure 14B]

[0147] 1 is an example of an AR display of a product in a user device application. [Figure 15A]

[0148] 1 is a flow chart of a process for tracking the ripeness of purchased produce. [Figure 15B]

[0149] 13 is an example display in a user device application for tracking the ripeness of purchased produce. [Figure 16]

[0150] 13 is an example display in a user device application for viewing the current ripeness of a scanned produce. [Figure 17]

[0151] FIG. 1 is a block diagram of system components that may be used to implement a system for assessing the quality of one or more food products. [Figure 18]

[0152] 1 is a flowchart of a process for training a model to assess product quality. [Figure 19A]

[0153] FIG. 1 is a conceptual diagram of an example cabinet that assesses quality metrics of products placed therein. [Figure 19B]

[0153] FIG. 1 is a conceptual diagram of an example cabinet that assesses quality metrics of products placed within it. [Figure 19C]

[0153] FIG. 1 is a conceptual diagram of an example cabinet that assesses quality metrics of products placed within it. [Figure 19D]

[0153] FIG. 1 is a conceptual diagram of an example cabinet that assesses quality metrics of products placed within it. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0125] Detailed Description

[0154] The present disclosure describes systems, methods and computer programs for assessing and dynamically identifying product metrics, such as metrics indicative of the quality of a product, and pricing the product accordingly. Product quality may be dynamically assessed in real time by utilizing stored data about the product and a product model. The disclosed technology may provide for a user to scan a product and receive information about the product and a product name from a computing system. A user may take an image of the product using a device such as a cell phone, laptop, camera and / or tablet. Sometimes the image may include a product without a label or other product identifier associated with the product. Sometimes the image may include a product with a label or other product identifier.

[0126]

[0155] For example, 1D / 2D barcodes may be used, included on stickers, printed directly on the product, and / or laser etched. The barcodes may be read by optical scanning devices such as cell phone cameras and cash registers. Color barcode systems may also be employed, which may use a geographical arrangement of colors to encode a unique identifier for the product. The unique identifier may be determined by optical recording and decoding.

[0127]

[0156] The information returned to the user may include an amount of time until the produce is ripe to eat. The information may include a current price for the produce, a discount based on the current quality of the produce, and / or a discount based on the expected quality of the produce in a particular amount of time. The price of the produce may be adjusted based on the particular or expected quality of the produce. For example, ripe or past-ripe produce may have a lower price than ripe or nearly ripe produce. This may motivate consumers to purchase the less expensive, lower quality produce first. As a result, lower quality (e.g., currently ripe or past-ripe) produce may be purchased, thereby reducing produce-based waste. This may also increase the perceived quality of the produce on the shelf and / or in the produce section. Additionally, because the information returned may indicate the current and / or expected quality of the produce, the user may not need to inspect the produce by hand or with their eyes before making a purchase decision. For example, the user may not need to crush the produce or take other destructive measurements to assess the produce quality.

[0128]

[0157] In some implementations, a retailer can place a non-destructive measurement system next to a produce container in a store. The device or system can constantly monitor the quality (e.g., freshness, ripeness, etc.) of the produce in the container and adjust the price of the produce accordingly. As a result, the produce may not need to be labeled or stamped with a produce identifier. Additionally, the non-destructive measurement system can be used to measure the days of remaining shelf life and / or quality, as well as print out and / or display coupons based on the system's determination. The system's determinations and measurements can also be used to train predictive quality and pricing models that can be used. Additionally, the non-destructive measurement system can be advantageous such that store employees do not need to collect and manage data. Labor costs can be reduced. Data collected by the system can be used to improve the accuracy and predictions of algorithms and models. In some implementations, the non-destructive measurement system can be implemented at multiple locations along the supply chain, such as distribution centers, storage facilities, etc. The non-destructive measurement system can include one or more sensors located around the produce container in the store. The sensors can be attached to the shelves holding the produce containers. The sensors can also be attached to the containers holding the produce. In some implementations, the container can be manufactured with sensors attached or integrated. The sensors can be configured to detect volatiles and other compounds released by the product. Signals collected by the sensors can then be transmitted to a computer system, which can assess the collected signals to identify quality metrics of the product in the container. The computer system can apply one or more machine learning trained models to the collected signals to assess the quality of the product. Based on the quality assessment, the computer system can determine one or more supply chain changes / adjustments as described herein.

[0129]

[0158] In some implementations, a grocer can scan a produce at a POS terminal. Based on information associated with the scanned produce, the POS terminal and / or a computing system in communication with the POS terminal can determine and / or adjust the price of the produce. Thus, the grocer can automatically offer the consumer a discount on the produce depending on the current quality of the produce. In some implementations, when a unique identifier (e.g., a label, a bar code, a QR code, an SKU, etc.) is scanned by a consumer (e.g., a shopper), the computing system can identify a coupon associated with the identifier, which can be scanned at a POS terminal or a cash register. Once scanned, the consumer can receive a discount on the produce at checkout. In some implementations, a printer can be present in the produce section that prints out coupons that may be redeemable for produce that have a particular uniquely identifying element. For example, the printer can print out coupons for produce that have stickers with a particular range of location identifiers.

[0130]

[0159] Referring to the figures, Fig. 1A is a conceptual diagram of automatic determination of produce quality. As shown in Fig. 1A, one or more computer systems, servers, and devices can be used to evaluate the produce 112A-N and determine produce quality metrics (e.g., ripeness metrics, freshness metrics, taste metrics, sweetness metrics) at different points during the supply chain lifecycle of the produce 112A-N. The evaluation system 102, the quality assessment system 104, the user devices 106A-N, and the data store 108 can communicate (e.g., wired and / or wireless) via a network 110. In some implementations, the systems 102 and 104 can be the same computing system or server.

[0131]

[0160] Some of the products 112A-N may include product identifiers 114A-N, such as labels, stickers, stamps, bar codes, SKUs, and QR codes. For example, some of the products 112A-N may be first scanned and labeled with a label 114A-N when they arrive at a distribution center in the supply chain. The label 114A-N may also be added at a packing facility as part of the process of packaging the products 112A-N. The label 114A-N may also be added after the products 112A-N are coated with a ripening agent / treatment. In some implementations, the products 112A-N may not be labeled or otherwise identified at the distribution center and / or before being delivered to a user at the user device 106A-N.

[0132]

[0161] The user devices 106A-N can be used by multiple different types of users, which may include, for example, consumers (e.g., shoppers), grocers, supermarket employees, chefs, restaurant employees, and other relevant stakeholders in the supply chain of the products 112A-N.

[0133]

[0162] The evaluation system 102 may be configured to perform an initial evaluation of the products 112A-N. The quality assessment system 104 may be configured to dynamically identify current quality information for the products 112A-N at a time later than when the evaluation system 102 performs the initial evaluation. The data store 108 may be configured to store data about the products 112A-N evaluated by the evaluation system 102. The data store 108 may be made up of multiple data stores. For example, as described throughout this disclosure, one or more data stores may be used to store information about the products 112A-N and models that predict product quality metrics. In some implementations, the data store 108 and any data stores described throughout this disclosure may be immutable ledgers. An immutable ledger may be advantageous for establishing trust with consumers and other relevant stakeholders in the supply chain.

[0134]

[0163] 1A, at time=1, the evaluation system 102 may evaluate the products 112A-N (step A) and label the products 112A-N with labels 114A-N (step B). Evaluating the products 112A-N (step A) may include generating information and / or predictive models for the products 112A-N. The generated information may be stored in the data store 108 and may later be used by the quality assessment system 104 to more accurately assess the quality of the products 112A-N.

[0135]

[0164] As an example, the evaluation system 102 may receive image data of the products 112A-N and, based on an evaluation of the image data, identify quality characteristics and / or quality scores of the products 112A-N represented in the image data. The image data may include hyperspectral images, associated metadata, x-rays, RGB models, etc. For example, the imaging devices may be configured to continuously capture image data of the products 112A-N as they move along a conveyor belt from one or more storage locations within the distribution center to other destinations within the distribution center. As another example, the image data may be captured by relevant stakeholders in the supply chain using devices such as user devices 106A-N.

[0136]

[0165] In some implementations, the evaluation system 102 may also receive destructive measurements of the products 112A-N. The destructive measurements may be modeled and correlated with non-destructive measurements, such as image data, to identify different quality characteristics of the products 112A-N. The destructive measurements may include penetrometer data and / or hardness meter data. The destructive measurements may be advantageous if the variability in the quality of the products 112A-N is sufficiently low.

[0137]

[0166] Models can be generated and trained using machine learning techniques (e.g., convolutional neural networks) to process image data and identify quality features and scores. Different quality features can be modeled and identified based on produce and / or produce type. Examples of quality features can be color, firmness, ripeness, size, shape, crispness, flavor, sweetness, moisture content, taste profile, likelihood of mold, nutrients, pesticide use, bruising, spoilage, and dryness.

[0138]

[0167] For example, when evaluating the produce 112A-N (step A), the evaluation system 102 may receive image data including histograms, RGB color models, hyperspectral data, multispectral data, etc. The image data may include images of a particular produce with some particular characteristics to model, such as spoilage and dryness, and images of the same type of produce without the particular characteristics to model. In some implementations, the image data may include images of the exterior and / or interior of the produce. In some implementations, the image data may include images of a particular produce at or between different stages of ripeness. The image data may be a robust collection of training data indicative of multiple different characteristics that may be present in the particular produce throughout the life cycle of the produce. The system 102 may then identify features of the produce 112A-N from the image data. Such features may be indicative of the quality of the produce 112A-N. The identified features may then be labeled. The system 102 may generate a machine learning trained model for each identified feature. The model can be trained to detect the features and assess the overall quality of the products 112A-N at a later time (e.g., when the products 112 are in a store and a consumer requests to know the current quality of one or more of the products 112A-N). The generated model can be stored in the data store 108 and later accessed by the quality assessment system 104.

[0139]

[0168] As an illustrative example, test image data of limes (e.g., produce 112A-N) may be received. Using segmentation and analysis techniques, blemishes may be a feature identified by the evaluation system 102 as an indicator of lime quality. Image data of limes that show signs of blemishes may be labeled as blemished, while image data of limes that do not show signs of blemishes may be labeled as good limes. A convolutional neural network (CNN) may be used to train a lime blemish model to distinguish image data that shows blemishes from image data that does not show blemishes. For example, the model may be trained to analyze each patch and / or pixel in the image data to see if lime is present and, if lime is present, to see if the lime appears to show signs of blemishes based on the labeled test image data. If the lime appears to show signs of blemishes, the model may be trained to tag or otherwise classify / label the lime in the image data as blemished, which may lead to the lime being classified as being of less good quality. The model may also be trained to tag limes with other descriptors for blemishes, including but not limited to numerical, Boolean and / or string values.

[0140]

[0169] Still referring to FIG. 1A, once the products 112A-N are evaluated (step A), the evaluation system 102 can classify the products 112A-N (step B). In other words, labels 114A-N can be generated on the products 112A-N and used to identify the products 112A-N. The labels 114A-N can then be attached to or associated with the products 112A-N. Classifying the products 112A-N can include identifying and referencing codes already placed on the products 112A-N or on the packaging of the products 112A-N. The evaluation system 102 can also transmit product data (e.g., information, models, characteristics, etc. identified when evaluating the products 112A-N in step A) to the quality assessment system 104 (step C). In some implementations, the quality assessment system 104 can store the product data in the data store 108 (step D).

[0141]

[0170] At time=2, the labeled products 112A-N may be delivered to a user (step E). For example, the products 112A-N may be delivered to a supermarket, a restaurant, a market, etc. Time=2 may be a later time than time=1. In some implementations, time=2 may be the same time as time=1.

[0142]

[0171] At time=3, one or more users at user devices 106A-N may request product quality information (step F). For example, as described further below, a consumer may scan the label 114A of the product 112A at a supermarket. Once the label 114A is scanned with the user device 106A, the user device 106A may send a request to the quality assessment system 104 to receive information about the current quality of the product 112A. Time=3 may be a time later than time=1, when the product 112A is first evaluated.

[0143]

[0172] At time=4, the quality assessment system 104 can identify the product 112A (step G) and determine the current quality of the product 112A (step I). Time=4 can be the same time as time=3. Time=4 is later than time=1. The quality assessment system 104 can use object recognition techniques to identify the product 112A. For example, a request for product quality metrics can include image data or other data captured by the user device 106A of the product 112A. Using object recognition techniques, the system 104 can separate the product 112A and / or the label 114A in the image data. The system 104 can then decode the label 114A to identify the product 112A. Using the decoded information, the system 104 can retrieve a model associated with the product 112A from the data store 108 (step H). The model was generated by the evaluation system 102 at time=1 and can be used to assess and predict the current quality of the product 112A.

[0144]

[0173] Using the retrieved model, the quality assessment system 104 can determine the current quality of the product 112A (Step I). The system 104 can also determine the current quality of the product 112A based on additional information that can be retrieved from the data store 108 and / or additional information received from one or more other computer systems, devices, servers and / or data stores. For example, the current quality can be further determined based on historical information of the product 112A throughout the supply chain, geographic information about where the product 112A was originally evaluated and / or where the product 112A was scanned by the user device 106A, and other supply chain information. As another example, the current quality of the product 112A or a batch of products 112A-N can be determined based on collecting and analyzing volatile substances and / or other compounds emitted by the product 112A. As described with reference to Figures 1B-1F, the volatile substances and other compounds emitted by the product 112A can be measured by a sensor located near the product 112A. The detected volatiles and compounds can be sent to the quality assessment system 104 for processing. The system 104 can, for example, apply one or more machine learning trained models to the detected volatiles and other compounds to identify the current quality of the produce 112A and predict the ripeness curve of the produce 112A and / or the future quality, ripeness or other characteristics of the produce 112A. As described above, the system 104 can search for one or more models corresponding to the produce 112A and / or the particular quality characteristic to be assessed and modeled. For example, the system 104 can search for an avocado ripeness model, an avocado taste model, an avocado firmness model and / or an avocado odor / aroma model. One or more other models specific to the type of produce 112A and / or the particular quality characteristic to be assessed and modeled can be searched. For further discussion on identifying the current quality of a produce based on the assessment of volatiles and other compounds emitted by the produce, see FIG. 1B.

[0145]

[0174] The system 104 can then transmit the current quality of the product 112A to the requesting user device 106A. As shown and described, the system 104 can dynamically process requests for product quality metrics at some later time after an initial evaluation of the requested product by the evaluation system 102. Thus, the quality assessment system 104 can provide the requesting user device 106A with current real-time quality information about the product 112A based on a robust set of data and models associated with the product 112A.

[0146]

[0175] 1B is a conceptual diagram of a produce display system 132 for identifying and presenting produce quality information to a user in a store 130. The system 132 includes produce containers 120A-N disposed on a support structure 123. The support structure 123 may be a shelf, rack, or another structure for supporting one or more produce containers 120A-N on a display shelf in the store 130. The support structure 123 may be movable such that the produce containers 120A-N can be moved and / or positioned on different displays / configurations in the store 130. In some implementations, the support structure 123 may be in a stationary position in the store 130 and the produce containers 120A-N can be swapped and / or moved to a different support structure 123 in the store 130.

[0147]

[0176] Produce bins 120A-N may be configured to contain one or more different types of produce 134A-N, as further described in Figures 1C-1F. For example, produce bin 120A contains oranges as produce 134A. Produce bin 120D contains green apples as produce 134D. As another example, produce bin 120G contains tomatoes as produce 134G. One or more other types of produce 134A-N may be included in produce bins 120A-N. Produce bins 134A-N may include, but are not limited to, apples, citrus fruits, berries, melons, peppers, tomatoes, leafy produce, fruits, vegetables, legumes, nuts, flowers, processed foods, candy, vitamins, nutritional supplements, and the like.

[0148]

[0177] The produce bins 120A-N can be any type of bin, container, crate, box, or other structure for holding a large quantity of produce 134A-N. The produce bins 120A-N can be configured for / attached to a support structure 123 of the produce display system 132. The produce bins 120A-N can also be placed in / on the support structure 123 and movable to other positions within the support structure 123 or to other support structures 123. For example, the produce bins 134A-N can be transported to a storage facility or from a farm to the store 130 within each produce bin 120A-N. These produce bins 120A-N can then be placed in / on the support structure 123 of the produce display system 132 when the produce bins 120A-N are transported to the store 130. Thus, the produce bins 134A-N can be moved through the supply chain within the same produce bins 120A-N.

[0149]

[0178] As further described with reference to FIGS. 1C-1F, each product container 120A-N may include an array of sensors 122A-N. The sensors 122A-N may be configured for and / or retrofitted to existing product containers 120A-N. In some implementations, the product containers 120A-N may be manufactured with an array of sensors 122A-N integrated therein. The sensors 122A-N may be configured to detect one or more different types of compounds or volatile substances released from the products 134A-N in each product container 120A-N. Each of the arrays of sensors 122A-N (or individual sensors) may be configured to detect different types of compounds and / or concentrations of compounds based on the type of product contained in each product container 120A-N. In some implementations, the array of sensors may have one or more sensors configured to detect each of the volatile substances or other compounds of interest. In an illustrative example, a produce container containing avocados may include sensors 122A-N configured to detect threshold amounts of total volatile organic compounds (TVOCs) and odors emitted by the avocados, while a produce container containing apples may include sensors 122A-N configured to detect different threshold amounts of esters, alcohols, aldehydes, terpenoids, or other types of compounds emitted by the apples.

[0150]

[0179] Some of the sensors 122A-N may be located above or around the product containers 120A-N. For example, the sensors 122A-N may be located a predetermined distance above the product containers 120A-N. The sensors may be located 1 to 6 inches away from the product container. As another example, the sensors may be located 2 inches above the product container, 5 inches above the product container, 8 inches above the product container, 10 inches above the product container, etc. In some implementations, multiple sensors may be located around the product container at predetermined distances from the product container. For example, the sensors may be located on each side of the product container and / or at a predetermined distance from the bottom of the product container and / or the top of the product container. One or more of the sensors 122A-N may also be located / mounted on the support structure 123 of the product display system 132. One or more of the sensors 122A-N may also be incorporated within the support structure 123 of the product display system 132 and / or the product containers 120A-N.

[0151]

[0180] Each produce container 120A-N also includes a respective display 136A-N for presenting information about the produce 134A-N within each produce container 120A-N. Examples of displays 136A-N include displays capable of receiving and presenting information electronically, such as a liquid crystal display (LCD) or a light emitting diode (LED) display. The displays 136A-N may be attached to each produce container 120A-N and may be dynamically updated to present current information about the produce 134A-N. For example, as described further herein, the displays 136A-N may present information such as the name of the produce 134A-N, discounts, prices, or offers associated with the produce 134A-N, and quality metrics associated with the produce 134A-N (e.g., whether the produce 134A-N is ripe or not, ready to eat or not, sweet or not, sour or not, ripe or not, juicy or not, etc.). In some implementations, one or more of the displays 136A-N may be mounted or otherwise configured to a support structure 123 of a product display system 132 described herein.

[0152]

[0181] Still referring to FIG. 1B, components of the product containers 120A-N, such as displays 136A-N and sensors 122A-N, are in electronic communication (e.g., wired and / or wireless) with the quality assessment system 104. The sensors 122A-N can collect volatiles or other compounds emitted by the products 134A-N in each product container 120A-N (Step A). ​​As described herein, each of the sensors 122A-N can be configured to detect different volatiles or other compounds. Groups of sensors can also be configured to detect the same volatiles of interest. Each group of sensors can detect a different volatile (or other compound) of interest.

[0153]

[0182] The volatile substances or compounds may be collected in batches and then transmitted to the quality assessment system 104 (Step B). In some implementations, the volatile substances or compounds may be collected in real time and transmitted to the system 104 (Step B). The system 104 may receive the volatile substances or other compounds data (Step B). The system 104 may also retrieve a product model from the data store 108 (Step C). The system 104 may determine which product model to retrieve based on the type of product 134A-N that corresponds to the received volatile substances or other compounds data. The system 104 may also determine which product model to retrieve using techniques described throughout this disclosure.

[0154]

[0183] The quality assessment system 104 can apply the retrieved models to the volatiles or other compound data to identify the quality of the produce 134A-N in the particular produce container 120A-N (step D). For example, the system 104 can receive TVOC data from sensors 122A and 122G in / around the produce container 120A. The TVOC data can correspond to oranges (produce 134A). The system 104 can search for one or more models in the data store 108 associated with identifying quality metrics for oranges. For example, the system 104 can search for a model that identifies the juiciness of oranges, another model that identifies the sweetness or sourness of oranges, and another model that identifies the shelf life of oranges. The system 104 can then identify characteristics / quality metrics such as juiciness, sweetness, and shelf life of oranges based on applying the retrieved models to the received TVOC data. The system 102 can identify quality metrics for all produce in the produce container.

[0155]

[0184] The quality assessment system 104 may then generate an output for each product receptacle 120A-N regarding the quality of the products 134A-N (step E). The output may be transmitted to a respective display 136A-N for the product receptacle 120A-N. The displays 136A-N may then display the output, such as the quality information, for each product 134A-N (step G). Because the displays 136A-N are individually controlled by the quality assessment system 104, the displays 136A-N may be updated independently of one another to reflect information about the particular product 134A-N contained in each product receptacle 120A-N.

[0156]

[0185] In some implementations, the output may be transmitted to one or more other devices and / or computing systems. For example, the output may be quality information about the produce 134A-N in each produce container 120A-N, which may be transmitted and presented to a mobile application on a customer's user device in the store 130. See FIGS. 13-16 for further discussion of quality information that may be presented on a user device.

[0157]

[0186] In some implementations, the output may include the quality of the products 134A-N in each product bin 120A-N and may be returned for use in supply chain change decisions (step F). For example, the output may be transmitted to the pricing system 202 shown in FIG. 2, which may use the output to determine price adjustments, discounts, and / or special offers for each product 134A-N. The output may also be transmitted to a point-of-sale terminal, such as the POS terminal 302 described with reference to FIG. 3. The POS terminal may then correlate the quality of each product 134A-N to price adjustments during the checkout process for an article that includes each product 134A-N. The output may be transmitted to one or more other computing systems and / or devices for use in supply chain changes, as described throughout this disclosure.

[0158]

[0187] Sometimes steps A-G may be performed for only some of the products 134A-N in some of the product containers 120A-N. For example, volatiles may be collected at different times or during different time intervals in each of the product containers 120A-N. The quality assessment system 104 may perform steps B-F for the products in which volatiles were collected during a first time period, and then perform steps B-F for the products in which volatiles were collected during a second time period. The system 104 may accordingly generate outputs presented on the displays 136A-N of the product containers 120A-N in which volatiles were collected and analyzed.

[0159]

[0188] In some implementations, the store 130 may include one or more gas chromatography machines or systems. Such machines or systems may be in communication with components of the produce containers 120A-N, such as the sensors 122A-N and displays 136A-N. The volatiles collected by the sensors 122A-N may be transmitted to a gas chromatography machine in the store 130, which may process the volatiles and generate volatile data. The volatile data may then be analyzed by a quality assessment system 104 that applies one or more machine learning trained models described herein. In some implementations, the volatiles may be transmitted to the system 104 as described above. The system 104 may then analyze the volatiles according to techniques described herein.

[0160]

[0189] FIG. 1C is a perspective view of a produce container 120A used to perform the techniques described herein. The produce container 120A can be used in combination with the machine learning techniques described herein to determine quality metrics for a batch of produce 134A contained in the produce container 120A. For example, the disclosed techniques can be used to determine an average quality score for the produce 134A. The average quality score can then be used to determine information about the quality of the produce 134A that is presented / output and dynamically updated on the display 136A of the produce container 120A. In some implementations, the average quality score can also be used to determine how to categorize the produce 134A within the produce container 120A and / or how to categorize the produce container 120A in relation to other produce containers in an in-store produce display system, as will be further described with reference to FIGS. 5A-5D.

[0161]

[0190] As shown and described herein, the product container 120A may include an array of sensors 122A-N. The sensors 122A-N may detect the presence and / or amount of one or more volatile substances or other compounds emitted by the product 134A. The product container 120A may optionally include a power source 138, a communication interface 140, and / or an outlet 142A-N. The power source 138 may be a battery. The power source 138 may also be a solar panel. The power source 138 may be rechargeable and / or replaceable. The power source 138 may provide a power supply to the components of the product container 120A, such as the sensors 122A-N, the display 136A, the communication interface 140, and / or the outlet 142A-N. The communication interface 140 may provide communication (e.g., wired and / or wireless) between the components of the product container 120A and a computer system (e.g., the quality assessment system 104) described throughout this specification. Outlets 142A-N may be configured to connect to an external power source, which can provide power to the components of product container 120A. One or more of components 122A-N, 136A, 138, 140, and 142A-N may be integrated or incorporated into product container 120A. One or more of components 122A-N, 136A, 138, 140, and 142A-N may also be incorporated into an existing product in-store or within the supply chain.

[0162]

[0191] The sensors 122A-N may include metal oxide sensors. In some implementations, the sensors 122A-N may include infrared sensors. The sensors 122A-N may also include gas chromatography sensors and / or photoionization detectors (PIDs). Generally, the PIDs emit light (e.g., ultraviolet light), which can split the compounds 150 into positive and negative ions. The positive and negative ions can be detected by the sensors 122A-N and can indicate the presence of the parent volatile compounds. The metal oxide sensors use a metal film to detect the compounds 150, which can include benzene, ethanol, toluene, etc. One or more other types of sensors can be used to collect different types of volatiles or other compounds released by different types of products described throughout this disclosure. For example, one or more of the sensors 122A-N can be an electronic nose using any combination of the sensors described herein (e.g., with reference to Figures 19A-19D). The electronic nose may have a combination of sensors including, but not limited to, nanocomposite sensors, electromechanical sensors, metal oxide semiconductors, photoionization sensors, and / or non-dispersive infrared sensors (NDIR). In some implementations, the sensors 122A-N may be any other type of sensor that is not part of the electronic nose, such as nanocomposite sensors, electromechanical sensors, metal oxide semiconductors, photoionization sensors, and / or NDIR sensors.

[0163]

[0192] The sensors 122A-N can be located along the walls (e.g., side walls, front wall, and / or back wall) of the product container 120A. The sensors 122A-N can also be located on the inside bottom of the product container 120A. The placement of multiple sensors 122A-N throughout the product container 120A can provide accurate collection of volatiles and other compounds released by the product 134A inside the product container 120A. For example, placement of the sensors 122A-N along the inside bottom of the product container 120A can be beneficial to ensure that the sensors 122A-N are in intimate contact with the product 134A and closer to the volatiles and other compounds released by the product 134A. As a result, the sensors 122A-N can accurately capture larger and more concentrated quantities of volatiles and other compounds released by the product 134A. In some implementations, a vacuum system can be used to draw air across the sensors 122A-N so that emitted volatiles or compounds are directed toward the sensors 122A-N. See Figures 19A-19D for further discussion of this configuration with a vacuum system.

[0164]

[0193] An array of sensors 122A-N may also be incorporated into the product container 120A to filter out certain volatiles or other compounds that may get into or around the product container 120A from other products in other product containers. Some of the sensors 122A-N may be located near the top of the product container 120A because some of the volatiles released by the product 134A may be in low concentrations and may rise. Some of the sensors 122A-N may be located near the bottom of the product container 120A because some of the volatiles released by the product 134A may be in high concentrations and therefore will not rise much, if at all.

[0165]

[0194] Some of the sensors 122A-N may be configured to detect and transmit volatiles and other compounds in real time to one or more computing systems described herein (e.g., quality assessment system 104). At times, the sensors 122A-N may collect volatiles until a predetermined / threshold amount of volatiles is collected. Once the predetermined / threshold amount of volatiles is collected, the collection may be transmitted to a computing system described herein. In some implementations, the volatiles may be collected on a material in or near the product container 120A. Once the predetermined / threshold amount of volatiles is collected on the material, the material is automatically heated to a predetermined temperature at a fixed time interval to release a population of volatiles to the sensors 122A-N. As a result, the sensors 122A-N may collect accurate readings of high concentrations of volatiles released by the product 134A in the product container 120A over a specific period of time. Such a technique may be referred to as heat absorption.

[0166]

[0195] Additionally, the power supply 138, communication interface 140 and / or outlets 142A-N may be configured on the back wall, side wall and / or bottom of the product container 120A so as not to interfere with the collection of volatiles and other compounds by the sensors 122A-N. The components 138, 140 and / or 142A-N may also be attached to or integrated into the product container 120A in a location that does not impede the storage capacity of the product container 120A or interfere with the simple and aesthetic look of the product container 120A in the store.

[0167]

[0196] Using machine learning trained models and disclosed techniques, the compounds 150 can be correlated to one or more freshness states of a batch of produce 134A contained in the produce container 120A. In some implementations, the compounds 150 can be correlated to other quality characteristics / metrics of the produce 134A, such as taste, ripeness, shelf life, sweetness, sourness, tartness, juiciness, fall, etc. of the produce 134A. Furthermore, the compounds 150 emitted by the produce 134A vary based on the type of the produce 134A. For example, an avocado can emit benzenes, acids, aldehydes, and / or ketones during its life cycle, while other produce, such as apples or oranges, can emit esters, alcohols, benzenes, and / or toluenes. As another example, ripe produce can emit high levels of esters, while unripe produce can emit lower levels of esters.

[0168]

[0197] The sensors 122A-N may be positioned and configured to detect different types of compounds based on the type of produce 134A contained in the produce container 120A. As an illustrative, non-limiting example, the ripeness of an avocado may be determined based on detection of one or more of the following aldehydes: acetaldehyde, hexanal, trans-2-hexen-1-ol, trans, trans-2,4-hexadienal, trans-2-nonenal, furfural, 2,4-heptadienal, and / or trans-2-decenal. As another illustrative, non-limiting example, the ripeness of an avocado may be determined based on detection of one or more of the following alcohols: ethanol, hexanol, pentanol, and / or pentenol. As another illustrative, non-limiting example, the ripeness of an avocado can be determined based on the detection of one or more of the following esters: methyl acetate, ethyl acetate, methyl hexanoate, ethyl propionate, isoamyl acetate, ethyl butyrate, heptyl acetate, 3-heptyl acetate, ethyl 4-methylvalerate, and / or ethyl acetoacetate. As another illustrative, non-limiting example, the ripeness of an avocado can be determined based on the detection of one or more of the following terpenes: limonene, myrcene, carvone, linalool, ionone, carveol, farnesol, pinene, guaiene, and / or cumene. As another illustrative, non-limiting example, avocado sclerotinia can be determined based on the detection of one or more of the following terpenes: limonene, myrcene, carvone, linalool, ionone, carveol, farnesol, pinene, guaiene, and / or cumene. Similar compounds can be detected in other foods (e.g., fruits, produce), including, but not limited to, mandarins, mangoes, and apples. Detection of such compounds in mandarins and mangoes can be used to identify off-flavors and / or fermentation quality of such foods (e.g., when the food is no longer fit for consumption), and detection of such compounds in apples can be used to identify quality differences across different varieties of apples.

[0169]

[0198] 19A-19D, the product container 120A may include one or more additional elements to direct a more concentrated amount of the compound 150 to the sensors 122A-N. These elements may include, but are not limited to, a cone, a probe, a fan, and / or a cap.

[0170]

[0199] As described herein, the compounds 150 detected by the sensors 122A-N can be transmitted via the communication interface 150 to a computing system, such as the quality assessment system 104 of FIG. 1A, for processing and analysis. In some implementations, the produce container 120A can include a controller 143 configured to receive the collected compounds 150 (or data about the collected compounds 150) from the sensors 122A-N and determine one or more quality metrics of the produce 134A in the produce container 120A based on the collected compounds 150. The quality metrics can include freshness parameter values ​​of the produce 134A, such as shelf life, ripeness, sweetness, sourness, flavor, juiciness, etc. The controller 143 of the produce container 120A or the quality assessment system 104 described herein can determine one or more supply chain changes based on the freshness parameter values. The supply chain changes may include pricing adjustments, price discounts, shelf life identification (e.g., current shelf life, remaining shelf life, etc.), ripeness curves (e.g., current ripeness, amount of time until ready to eat, etc.), and / or shelf and / or container classification / arrangement suggestions. The controller 143 transmits one or more of the supply chain changes to a display device, such as the display 136A. The controller 143 may also transmit the supply chain changes to a user device, such as the user devices 106A-N of a shopper in the store. Thus, when a shopper uses their device to look up or scan the produce 134A in the store, information indicative of the freshness parameter values ​​of the produce 134A may be presented in a mobile application on the device. As another example, the price adjustments, discounts, remaining shelf life, and / or ripeness curves identified for the produce 134A may be presented in a mobile application on the shopper's device. For further discussion, see FIGS. 2 and 13-16.

[0171]

[0200] FIG 1D is a side view of the product container 120A of FIG 1C. The product container 120A has a front wall 144, a rear wall 146 opposite the front wall 144, and a first side wall 148 connecting a first edge of the front wall 144 and a first edge of the rear wall 146. In the example of FIG 1D, the sensors 122D and 122E can be configured, attached and / or incorporated inside the first side wall 148. Thus, the sensors 122D and 122E can be in intimate contact with the product and / or compounds released by the product contained in the product container 120A. One or more additional or fewer sensors can be attached inside the first side wall 148. Additionally, one or more of the sensors may be mounted closer to the top of the first sidewall 148, the bottom of the first sidewall 148, the sides of the first sidewall 148, and / or the central area of ​​the first sidewall 148. The same or similarly positioned sensors may be configured on an opposing second sidewall (not shown) of the product container 120A.

[0172]

[0201] The display 136A can be attached to an exterior surface of the front wall 144. The display 136A can be integrated into the front wall 144. In some implementations, the display 136A can be incorporated into an existing produce container 120A using fastening elements, such as nails, screws, bolts, adhesives, and / or magnets.

[0173]

[0202] The at least one sensor 122F can be attached to and / or integrated into the inner surface of the front wall 144. The at least one sensor 122F can be located anywhere along the length and / or height of the inner surface of the front wall 144. In some implementations, the at least one sensor 122F can be part of the display 136A or otherwise integrated with the display 136A.

[0174]

[0203] At least one sensor 122N can be attached to and / or incorporated into the inner surface of the rear wall 146. At least one sensor 122N can be located anywhere along the length and / or height of the inner surface of the rear wall 146. Additionally, the power source 138 and / or the communication interface 140 can be attached to or incorporated into the rear wall 146 of the product container 120A. Although not shown in FIG. 1D, the outlets 142A-N and / or the controller 143 described in FIG. 1C can also be attached to or incorporated into the rear wall 146 of the product container 120A. Any of the components 138, 140, 142A-N and / or 143 can be configured on the inner and / or outer surface of the rear wall 146. Any of components 138, 140, 142A-N and / or 143 may be integrated into or otherwise attached to the inner or outer bottom of product container 120A, thereby not interfering with the use of the interior of product container 120A to contain product. Thus, any of components 138, 140, 142A-N and / or 143 are configured on product container 120A so as not to interfere with the use of product container 120A as a container for product and accurate collection of compounds released by the product.

[0175]

[0204] As previously mentioned, the sensors 122A-N can be incorporated into the walls 144, 146, and 148 of the product container 120A. The sensors 122A-N can also be attached to the walls 144, 146, and 148 using fastening elements such as nails, screws, bolts, adhesives, and / or magnets.

[0176]

[0205] FIG. 1E is a front view of the product container 120A of FIG. 1C. As shown in this figure, the display 136A is attached to the outer surface of the front wall 144. The sensors 122F, 122J, 122K, and 122L can be attached to the inner surface of the front wall 144. Any of the sensors 122F, 122J, 122K, and 122L can be attached to different locations on the inner surface of the front wall 144. For example, attaching one or more sensors closer to the top of the inner surface of the front wall 144 can be beneficial to identify and distinguish compounds released by the product in the product container 120A from compounds in the surrounding environment. One or more additional or fewer sensors can be attached to the front wall 144. Sometimes, one or more sensors can be part of the display 136A. Thus, fewer sensors can be attached to, configured in, or incorporated into the front wall 144.

[0177]

[0206] 1F is a top view of the produce container 120A of FIG. 1C. As shown and described herein, the produce container 120A includes a front wall 144 to which the display 136A is attached, a rear wall 146, a first side wall 148, a second side wall 150 opposite the first side wall, and a bottom 152. At least sensors 122T and 122U can be attached to the first and second side walls 148 and 150, respectively. One or more sensors can be attached to the inner surfaces of the first and second side walls 148 and 150, the front wall 144, and the rear wall 146. In some implementations, one or more sensors, such as sensor 122U, can be attached to the outer surfaces of the first and second side walls 148 and 150, the front wall 144, and / or the rear wall 146. Mounting sensors on the exterior of the product container 120A can be beneficial to distinguish between compounds released by the product inside the product container 120A and compounds floating in the surrounding environment, thus detecting a more accurate reading of compounds released by the product inside the product container 120A and using it to identify product quality and / or determine changes to the product supply chain.

[0178]

[0207] 1F, sensors 122M, 122N, 122O, 122S, 122R, 122Q, and 122P can be attached to the bottom 152 of the product container 120A. One or more additional or fewer sensors can be configured at the bottom 152 of the product container 120A. The placement of sensors at or near the bottom 152 of the product container 120A can be beneficial for more accurate and / or higher concentration collection of compounds released by the product contained in the product container 120A.

[0179]

[0208] Although Figures 1C-1F are described with reference to product container 120A, the same or similar configurations are applicable to all product containers 120A-N described throughout this disclosure.

[0180]

[0209] FIG. 2 is a conceptual diagram of dynamic determination of product pricing metrics (e.g., price, discount, markup, pricing category). As shown in FIG. 2, one or more computer systems, servers, and devices can be used to identify current product quality metrics and adjust the price of the product based on the identified quality. The quality assessment system 104, the user devices 106A-N, the pricing system 202, the product model data store 116, the product data store 118, and the pricing model data store 204 can communicate (e.g., wired and / or wireless) via the network 110. In some implementations, the systems 104 and 202 can be the same computing system or server. In some implementations, one or more of the data stores 116, 118, and 204 can be the same data store (e.g., data store 108 of FIG. 1A), database, server, immutable ledger, and / or cloud storage.

[0181]

[0210] A user (such as a shopper, grocer, store employee, and / or restaurant / food service employee) at one of the user devices 106A-N, e.g., user device 106A, may scan a produce (step A). ​​User device 106A may request produce quality information from the quality assessment system 104 (step B). As described throughout, the request may include information such as scan data collected by user device 106A. The information may include one or more images of the scanned produce, a produce identifier associated with the produce, and any other information that user device 106A may scan or otherwise collect (e.g., as input from a user) about the produce.

[0182]

[0211] Using the information in the request, the quality assessment system 104 can identify the product (step C). As described herein, the quality assessment system 104 can use object recognition techniques to extract an image of the product from information that may include image data scanned / captured by the user device 106A. The quality assessment system 104 can also use object recognition or other image analysis techniques to identify the product identifier and decode the product identifier.

[0183]

[0212] Once the produce is identified, the quality assessment system 104 may use the produce identifier to retrieve data associated with the produce from the produce data store 118 (step D). The produce data may include information such as historical ripeness information, storage conditions, production location, historical quality information, quality characteristics, etc. The quality assessment system 104 may also use the produce identifier to retrieve a quality assessment model associated with the produce from the produce model data store 116 (step F). The model may be trained using machine learning techniques to detect one or more quality characteristics in the information received in the request. The model may also be configured to determine an overall quality of the produce based on the produce data and the information received in the request.

[0184]

[0213] The quality assessment system 104 may apply the model to the information in the request (step F). For example, if the information includes image data of a produce, a model may be trained to identify the produce within a bounding box in the image data. Another model may be trained to perform object recognition techniques on the bounding box portion of the image data to identify one or more quality features of the produce. In some implementations, different models may be applied to identify different quality features. Thus, each model may be trained to identify a different quality feature. For example, one model may be trained to identify browning of the skin of the produce based on color data. Another model may be trained to identify spoilage or blemishes based on color data of the interior and / or exterior of the produce. Another model may be trained to identify firmness based on texture. Another model may be trained to identify taste based on color data. One or more additional models may be trained to identify one or more other quality features of the produce. Another model may be trained to identify the overall quality of the produce based on features identified by the previous models. Additional or fewer models may be applicable to identify the quality of the produce. Furthermore, the models can be applied successively, which may be advantageous for developing more robust and accurate quality analyses.

[0185]

[0214] The quality assessment system 104 can determine the current overall quality of the produce (step G). As described throughout this disclosure, the quality assessment system 104 can use one or more machine learning trained models and the retrieved produce data to determine the current quality of the produce. Thus, the quality assessment system 104 can determine the quality of the produce when the user scans the produce (step A) and a request for produce quality metrics is sent to the quality assessment system (step B). The quality assessment system 104 can also determine the overall quality of the produce, the ripeness curve of the produce and / or other freshness parameter values / quality metrics of the produce based on the collection and analysis / modeling of volatile substances and compounds emitted by the produce. For further discussion on determining quality metrics of the produce based on volatile substances, see Figures 1B-1F.

[0186]

[0215] The identified produce quality metric may be a numeric, Boolean, and / or string value. For example, the identified produce quality metric may be one or more of the following descriptors: "currently ripe," "approaching maturity," "overripe," "not yet ripe," etc. One or more other descriptors and / or values ​​may be used to indicate the identified produce quality metric. In some implementations, the quality metric may be a ripeness curve of the produce, as described herein. The ripeness curve may indicate the past, current, and expected / future ripeness of the produce. For further discussion, see FIG. 7.

[0187]

[0216] Additionally, as described throughout this disclosure, the determined produce quality metrics may also include additional information, which may be determined by the quality assessment system 104 using one or more machine learning trained models, and may include the expected remaining shelf life of the produce, the expected time remaining until it is ready to eat, the expected time remaining until the produce is no longer consumable, firmness, tenderness, taste, nutritional content, suggested recipes based on the current quality of the produce, and suggested recipes based on the predicted quality of the produce.

[0188]

[0217] The quality assessment system 104 may store the identified product quality metrics in the product data store 118 (Step H). This information may be used to train and / or improve machine learning trained models and other algorithms that may be used by the quality assessment system 104 in identifying product quality metrics.

[0189]

[0218] The quality assessment system 104 may also transmit the determined product quality metrics to the pricing system 202 (step I). In this data packet, the quality assessment system 104 may transmit additional information, such as a product identifier associated with the product and one or more of the product data retrieved in step D. The quality assessment system 104 may also transmit the determined product quality metrics to the user device 106A.

[0190]

[0219] The pricing system 202 may be configured to determine a price adjustment and / or discount based on the current quality of the product. In some implementations, the pricing system 202 may also determine a price adjustment and / or discount based on the expected quality of the product. The pricing system 202 may retrieve a product pricing model from a pricing model data store 204 (step J). The system 202 may use information in the data packet received in step I to determine which pricing model to retrieve. In some implementations, the system 202 may also retrieve pricing models specific to the product as well as the user and / or relevant stakeholders in the supply chain. For example, a supermarket where the product is sold may have specific pricing models that can be used for different categories of products sold in the store. Thus, the system 202 may retrieve one or more pricing models associated with the supermarket.

[0191]

[0220] The pricing system 202 may apply the models (step K). A supermarket pricing model may, for example, be trained to reduce the price of a produce by a certain percentage when the produce is ripe to eat. One or more pricing models may be trained to reduce the price based on a decrease in the quality of the produce. One or more pricing models may be trained to increase the price based on an increase in the quality of the produce. One or more pricing models may also be trained to implement a logarithmic reduction in price once the produce is past its ripeness. One or more pricing models may be trained to cause the price to plummet once the produce is no longer ripe.

[0192]

[0221] Based on the applied model, the pricing system 202 can determine a price adjustment for the produce (step L). The price adjustment can also be determined based on modeling of volatiles released by the produce. For example, some volatiles released by the produce can be indicative of the sweetness, sourness, juiciness, flavor, etc. of the produce. The model can be applied to the volatiles to identify such characteristics for the produce. The pricing system 202 can then adjust the price of the produce based on whether the sweetness, sourness, juiciness, flavor, etc. of the produce meets a desired level or threshold for that produce. As an illustrative example, the volatiles released by a container of oranges can be modeled using a juiciness model. The model output can indicate that the container of oranges is dry rather than juicy. The pricing system 202 can use this identification to reduce the price of the oranges since the oranges are not at a level of juiciness desired by the consumer. As another example, the volatiles released by a container of apples can be modeled using a sourness model. The model output can indicate that the container of apples is very sour / tart rather than sweet. Using this identification, the pricing system 202 can decrease the price of the apples because they do not have the level of flavor desired by the consumer. As yet another example, the volatiles released by another container of apples can be modeled using a crispness model. The model output can indicate that this container of apples has an ideal level of crispness. Thus, the pricing system 202 can decide to increase the price of the apples because they have the level of crispness desired by the consumer. The pricing system 202 can also adjust the price of the apples based on one or more other quality attributes of the produce that can be identified by modeling the volatiles released by the produce. Such quality attributes can include, but are not limited to, citrus level, fermentation, flavor, sweetness, and overall taste. As described herein, models can be generated and used for each type of produce and for each quality attribute of a particular produce type.

[0193]

[0222] As mentioned, a price adjustment can be an increase or decrease in price based on the current and / or expected quality of the produce. The price can decrease as the quality of the produce decreases according to a curve that can describe the price elasticity of the quality of the produce. The price adjustment can also include a discount that can be applied to the price of the produce at the present time. The discount can be applied to the price of the produce later (e.g., after the produce is no longer ripe). In some implementations, less ripe produce can have a higher price than more ripe produce, thereby incentivizing consumers to purchase less ripe produce first. Doing so can reduce waste in the produce base.

[0194]

[0223] In step L, the pricing system 202 may also determine or predict the amount of time until the next price reduction. This determination may be based on the expected quality information received in step I from the quality assessment system 104 and / or retrieved from the produce data store 118. For example, if the produce is expected to be ripe for eating in the next two days, the pricing system 202 may determine that the price can be reduced after the next two days. The system 202 may also use techniques described herein to determine how much to reduce the price by after the next two days.

[0195]

[0224] The pricing system 202 can return the price adjustment to the user device 106A (step M). The pricing system 202 can also transmit the product quality metric to the user device 106A. In some implementations, the quality assessment system 104 can transmit the product quality metric to the user device 106A (step I). Sometimes, the identified product quality metric can be transmitted to the user device 106A before the price adjustment is transmitted to the user device 106A. Sometimes, the identified product quality metric and the price adjustment can be transmitted to the user device 106A simultaneously.

[0196]

[0225] The user device 106A can then output the product quality metrics and the price adjustment (see, e.g., FIGS. 13-16). This information can be presented on a graphical user interface (GUI) display of the user device 106A. For example, if the user device 106A is a shopper's mobile phone, the product quality metrics and the price adjustment can be output to a mobile application (see, e.g., FIGS. 4, 16). If the user device 106A is a store point of sale (POS) terminal at a checkout lane, the product quality metrics and the price adjustment can be output to the POS terminal (see, e.g., FIG. 3). The price adjustment can then be automatically applied to the transaction at the POS terminal. Thus, in some implementations, the user device 106A can also implement the price adjustment. As another example, the price adjustment can be presented to the shopper's user device 106A in the form of a coupon or discount barcode. The shopper can select a coupon or discount barcode on their device 106A and then present the barcode at checkout to have the price adjustment automatically applied.

[0197]

[0226] 3 is a conceptual diagram of dynamic adjustment of produce pricing metrics based on produce quality metrics. In this example, a grocer 304 is operating a POS terminal 302. A shopper may approach the grocer 304 with a produce 112A. The produce 112A includes a label 114A, which is a produce identifier. The grocer 304 may scan the produce label 114A at the POS terminal 302 (Step A). ​​For example, the label 114A may be scanned using a barcode scanner or other scanning device at the POS terminal 302. In some implementations, the grocer 304 may scan the label 114A by entering the value on the label 114A into the POS terminal 302.

[0198]

[0227] The POS terminal 302 may transmit the scanned label 114A to the quality assessment system 104 (step B). The quality assessment system 104 may proceed to retrieve product data from the product data store 118 (step C) and retrieve a product model from the product model data store 116 (step D) as described throughout this disclosure (see, e.g., FIG. 1A). The system 104 may determine current product quality metrics (step E) and transmit the current product quality metrics back to the POS terminal 302 (step F).

[0199]

[0228] In the example of FIG. 3, the POS terminal 302 may be configured to determine a pricing metric adjustment for the product 112A. Thus, the POS terminal 302 may be the same as or similar to the pricing system 202 of FIG. 2 and may perform the same or similar operations as the pricing system 202 (see, for example, FIG. 2, FIG. 11). Thus, upon receiving the current product quality metric in step F, the POS terminal 302 may retrieve product pricing information from the pricing model data store 204 and the product data store 118 (step G). As described throughout, the pricing information may include pricing models specific to the product 112A, the supermarket, or other relevant stakeholders in the supply chain. Using the retrieved information, the POS terminal 302 may correlate the current product quality metric with a price adjustment (step H). The POS terminal 302 may then output and / or implement the price adjustment (step I). For example, the price of product 112A may be automatically discounted and reflected in the output display of POS terminal 302 and on the shopper's receipt.

[0200]

[0229] FIG. 4 is a conceptual diagram of the identification of produce quality metrics to a consumer, such as a shopper. As shown, a shopper 402 can scan a produce 112A using a user device 106A (step A). ​​Scanning the produce 112A can include capturing an image of the label 114A using a camera of the user device 106A. For example, the shopper can capture an image of an avocado using a mobile device. The current quality of the avocado can be determined based on the captured image of the avocado, as described herein. The user device 106A can then transmit the scanned label 114A to the quality assessment system 104 (step B). As described throughout this disclosure, the system 104 can retrieve produce data from the produce data store 118 (step C) and retrieve a produce model from the produce model data store 116 (step D). Thus, the system 104 can identify the current produce quality metrics of the produce 112A (step E). The system 104 may transmit the current product quality metrics to the user device 106A and to the pricing system 202 (step F).

[0201]

[0230] The pricing system 202 may retrieve product pricing information from the pricing model data store 204 (step G). The system 202 may then correlate the current product quality metrics with the retrieved pricing information (step H). Based on such correlation, the pricing system 202 may determine a price adjustment for the product 112A (step I). The pricing system 202 may then transmit the price adjustment to the user device 106A (step J). The user device 106A may output the current product quality metrics and the price adjustment for display to the shopper 402 (step K).

[0202]

[0231] Based on the output information, the shopper 402 can make an informed decision as to whether or not to purchase the produce 112A. Furthermore, the shopper 402 may not need to use his or her own value judgment to determine whether the produce 112A is of a desirable quality. For example, the shopper 402 may not need to use destructive techniques, such as crushing the produce 112A, to determine whether the produce 112A is ripe and / or hard enough. Furthermore, the shopper 402 may be motivated to purchase the produce 112A at a lower price, where a lower price may indicate that the quality of the produce 112A is reduced or lower than other produce. Purchasing the produce 112A at a lower price may avoid produce-based waste and increase the overall quality of the produce on the shelf or in the produce section of the store.

[0203]

[0232] 5A is a conceptual diagram of store employee placement of shelves based on produce quality metrics. Employee 502 can use user device 106A to determine how to place produce 112A-N on display 504. Display 504 can include multiple shelves, such as shelf 506. Employee 502 can be tasked with replenishing shelf 506 with produce 112A-N. To avoid or reduce produce-based waste, it may be preferable to replenishing lower quality produce on shelf 506 before higher quality produce. As a result, shoppers can purchase lower quality produce before the lower quality produce is no longer consumable. In that case, the lower quality produce may not go to waste.

[0204]

[0233] As shown in FIG. 5A, an employee 502 can use a user device 106A to scan the labels of the products 112A-N (step A). ​​The employee can, for example, take a photo of the product 112A-N and use the photo to identify the current quality of the product 112A-N. The user device 106A can transmit the scanned label to the quality assessment system 104 (step B). The quality assessment system 104 can then transmit the identified product quality information back to the user device 106A (see, e.g., FIG. 1A) (step C).

[0205]

[0234] The user device 106A may output quality information associated with the products 112A-N (step D). As an illustrative example, the produce quality output 508 may include a table showing the ripeness level and number of days remaining until eating of each of the products 112A-N. In the example of Figures 5A and 5B, the product 112A may have a ripeness level of "not yet ripe" and 7 days remaining until eating. The product 112B may have a ripeness level of "ripe" and 4 days remaining until eating. The product 112C may have a ripeness level of "ripe" and 2 days remaining until eating.

[0206]

[0235] Based on the output 508, the user device 106A can determine a sorting order for the products 112A-N (step E). In some implementations, the user device 106A and / or the quality assessment system 104 can automatically determine how to sort the products 112A-N based on the current quality information. In some implementations, the employee 502 can determine how to sort the products 112A-N based on the outputted information.

[0207]

[0236] Employee 502 can then proceed to sort products 112A-N on shelves 506 (step F).

[0208]

[0237] FIG. 5B shows the display shelf 506 of FIG. 5A arranged by a store employee 502. The produce 112N is arranged at the front 510 of the shelf, while the produce 112A is arranged at the back of the shelf 512. The produce 112B is arranged between the produce 112N and the produce 112A. This classification of the produce 112A-N was determined based on the produce quality output 508. It may be desirable to arrange the produce that may be spoiled or of lower quality at the front 510 of the shelf 506, while the produce that is still not ripe or of higher quality is arranged at the back 512 of the shelf 506. Thus, as described throughout this disclosure, shoppers may be incentivized to purchase lower quality produce ahead of higher quality produce, thereby reducing or avoiding produce-based waste. Additionally, prices may be adjusted for each of the produce 112A-N based on their current quality, thereby further incentivizing shoppers to purchase lower priced, lower quality produce ahead of higher priced, higher quality produce. When a lower quality product (e.g., product 112N) is purchased, the higher quality products (e.g., products 112B and 112A) can be positioned near the front 510 of the shelf 506. The shelf 506 can also be replenished by adding higher quality products to the back 512 of the shelf 506.

[0209]

[0238] 5A and 5B, in some implementations, the quality of the products 112A-N may change daily. Thus, employee 502 may sort the products 112A-N such that whenever the quality of the products 112A-N changes, the lower quality products may always be presented at the front 510 of the shelf 506, while the higher quality products may always be positioned at the back 512 of the shelf 506.

[0210]

[0239] FIG. 5C shows a conceptual diagram of classification of the produce containers 120A-C based on the quality assessment of the products 134A-C therein. Each sensor 122A-N can collect measurements of volatiles 150A-N in each of the produce containers 120A-C (step A). ​​The volatile measurements can be transmitted to the quality assessment system 104, which can model the volatile measurements into a produce quality curve for each container (step B). In other words, the system 104 can determine a quality (e.g., ripeness) curve for the product 134A in the produce container 120A based on the volatiles 150A-N emitted by the product 134A, determine another curve for the product 134B in the produce container 120B based on the volatiles 150A-N emitted by the product 134B, and determine another curve for the product 134C in the produce container 120C based on the volatiles 150A-N emitted by the product 134B. For further discussion of generating the produce quality curves, see FIG. 7. Based on each quality curve, the quality assessment system 104 can identify produce quality metrics for each container (step C). For example, the system 104 can identify a current ripeness level for each batch of produce 134A, 134B, and 134C from each quality curve. The current ripeness level or other identified quality levels (e.g., remaining shelf life, taste, flavor, sweetness, crispness, sourness, odor, hardness, etc.) can be used by the system 104 to determine a suggested container sequence and display output (step D). As a result, the produce containers 120A-C can be sorted / arranged in order in the produce display system (e.g., see FIG. 1B) based on the identified quality of each product 134A-C.

[0211]

[0240] In the example of Figure 5C, the produce containers 120A-C may be sorted from most ripe produce (e.g., ready to eat) to least ripe produce (e.g., not yet edible). The produce containers 120A-C may also be sorted based on other metrics including, but not limited to, sweetest to least sweet, juicy to less juicy, less sour to more sour, less crunchy to crunchy, etc.

[0212]

[0241] Here, from left to right, produce container 120C with Gala apples 134C is initially in position 520, produce container 120A with Honeycrisp apples 134A is initially in position 522, and produce container 120B with Fuji apples 134B is initially in position 524. Once quality assessment system 104 has performed steps A-D, system 104 can determine a suggested new ordering of produce containers 120A-C based on the quality of the products 134A-N. In the new arrangement, produce container 120A with Honeycrisp apples 134A can be placed in position 520, produce container 120B with Fuji apples 134B can be placed in position 522, and produce container 120C with Gala apples 134C can be placed in position 524.

[0213]

[0242] Additionally, the system 104 may identify qualities of the produce 134A-C and generate updated outputs presented on the displays 136A-C of the produce containers 120A-N based on the identified qualities of the produce 134A-C. As shown in FIG. 5C, the display 136A of the produce container 120A may be updated with text indicating that the Honeycrisp apples are ready to eat. The display 136B of the produce container 120B may be updated with text indicating that the Fuji apples will be ready to eat in 2 days. The display 136C of the produce container 120B may be updated with text indicating that the Gala apples will be ready to eat in 3.5 days. The text output on each display 136A-C may be independently dynamically updated / changed as the current and / or expected quality of each produce 134A-C changes over time. In some implementations, the displays 136A-C can display pricing information based on the identified qualities of the products 134A-C in each product container 120A-C, as illustrated in FIG. 5D.

[0214]

[0243] FIG. 5D is a conceptual diagram of determining produce pricing based on a quality assessment of the produce 134A-C in each container 120A-N. Here, the displays 136A-C of each produce container 120A-C can be updated to reflect a price adjustment based on the identified quality of the produce 134A-C. The price adjustment can be based on various quality characteristics / metrics identified for the produce 134A-C, as described in connection with FIG. 2. In the example of FIG. 5D, the price adjustment is determined based on the current ripeness of the produce 134A-C. The more ripe the produce, the greater the price discount. The discount can motivate consumers to purchase ripe produce now, thereby avoiding or reducing produce-based waste. The price adjustment can also be determined based on one or more other quality metrics described herein, including, but not limited to, taste, odor, firmness, juiciness, crispness, and other metrics specific to a particular type of produce.

[0215]

[0244] As described with reference to FIG. 5C, measurements of volatile materials 150A-N may be collected at each container 120A-C using respective sensors 122A-N (Step A). ​​The volatile material measurements may be transmitted to a quality assessment system 104 which may model the volatile material measurements into product quality curves for each container (Step B). The system 104 may identify product quality metrics for each container based on the respective curves (Step C). The quality metrics may be transmitted to a pricing system 202 which may generate price adjustments and output displays 136A-C for each container based on the respective quality metrics (Step D).

[0216]

[0245] Here, produce container 120C containing Gala apples 134C is at or remains at position 520, produce container 120A containing Honeycrisp apples 134A is at position 522, and produce container 120B containing Fuji apples 134B is at position 524. Based on steps B-D performed by quality assessment system 104 and pricing system 202, pricing system 202 transmits pricing information which is output to each of displays 136A-C of produce containers 120A-C at their respective positions 520, 522 and 524.

[0217]

[0246] In the example of FIG. 5D, the Gala apples 134C in the produce container 120C are ready to eat (as determined by the system 104 in steps B-C), and therefore the display 136C is updated to include text indicating that the Gala apples 134C are discounted and ready to eat. The text may also include a discounted price for the apples 134C in the produce container 120C. Because the Honeycrisp apples 134A are nearly / almost ready to eat, the display 136A is updated to include text indicating that the Honeycrisp apples 134A are nearly / almost ready to eat. The text may include an adjusted price for the apples 134A, which may be higher than the Gala apples 134C, because the Honeycrisp apples 134A are not as ready / almost ready to eat as the Gala apples 134C. Because Fuji apples 134B are not yet ready to be eaten, display 136B is updated to include text reflecting the current ripeness / readiness level of Fuji apples 134B. The text may also include an adjusted price for Fuji apples 134B, which may be higher than Gala apples 134C and Honeycrisp apples 134A, because Fuji apples 134B are not yet at their peak / ideal for consumption by consumers.

[0218]

[0247] FIG. 6 is a conceptual diagram of produce identification and labeling. For example, when produce is shipped from a farm to a distribution center and / or packing facility, the produce may not yet be classified, labeled, or otherwise identified. Once at the distribution center or packing facility, the produce may be labeled so that it can be identified further down the supply chain. As described herein, the produce may be classified or labeled as part of the packing process. The produce may also be classified or labeled as part of coating the produce in a ripening agent treatment. Additionally, information about the produce, such as storage status, current quality, current ripeness, and production location, may be stored and associated with the produce via the produce's label (e.g., produce identifier, barcode, QR code, SKU).

[0219]

[0248] A supply chain worker 602 may scan the produce 112A using a user device 106A (Step A). ​​Scanning the produce 112A may include taking an image of the produce 112A with a camera of the user device 106A. In some implementations, the imaging device may be located at a distribution center and configured to automatically capture image data of the produce 112 as it moves into and throughout the distribution center.

[0220]

[0249] The supply chain worker 602 can be located at a distribution center or other storage facility. For example, the worker 602 can be tasked with scanning produce as it enters the distribution center. The supply chain worker 602 can be located anywhere else along the supply chain, such as on a farm, before or during the transition to a distribution center, and / or before or during the transition to a supermarket or other end consumer.

[0221]

[0250] The user device 106A can transmit the scanned data to the evaluation system 102 (step B). As an example, the scanned data can be image data. The image data can include hyperspectral data, associated metadata, and / or color models. The evaluation system 102 can also receive supply chain information at the current location of the product 112A (step C). The supply chain information can include the storage status of the product 112A, the amount of time the product 112A has been in transition, the amount of time the product 112A is expected to be stored at a distribution center before being shipped to an end consumer, what the end consumer expects upon receipt of the product 112A, the amount of transition time between the distribution center and the end consumer, information about the end consumer, and whether the product 112A underwent any ripening treatments or coatings before arriving at the current location.

[0222]

[0251] The evaluation system 102 may retrieve a product identification model from the product model data store 116 (step D). The identification model may be trained using machine learning techniques to identify the type of product from the scan data. For example, one or more of the models may be trained to extract the product 112A from the image data and identify feature points in the extracted portion of the image data that can be used to identify the product 112A. Once the feature points are extracted, one or more models may be used to correlate the feature points to the type of product. Thus, the evaluation system 102 may apply the model (step E) to identify the product 112A (step F). The product 112A may also be identified using the supply chain information received in step C. For example, if the supply chain information indicates the farm / location of production, the evaluation system 102 may use that information to eliminate possible matches between the scanned product 112A and the product type that the model was trained to identify. As an illustrative example, if a farm only produces peppers and green beans, it may eliminate possible product type matches with cucumbers, avocados, and bananas.

[0223]

[0252] Once the evaluation system 102 has identified the product 112A in step F, it may generate a label for the product 112A (step G). The label may be a unique identifier that may be used to associate any information about the product 112A with the product 112A. The label may then be stored in the product data store 118 (step H). Any additional information about the product 112A identified using the model may also be stored in the product data store 118.

[0224]

[0253] The evaluation system 102 can then send the label to the user device 106A (step I). The user device 106A can print the label (step J). The label can then be attached to the product 112A (step K). In some implementations, the user device 106A can communicate with an automated machine that automatically attaches the label to the product 112A. In some implementations, a supply chain worker 602 can attach the label to the product 112A. Once the product 112A is identified and labeled, the product 112A can be moved up the supply chain (step L). For example, the product 112A can be moved to storage. The product 112A can be moved to an end consumer, such as a supermarket or restaurant. The product 112A can also be moved to a food processing plant or any other relevant stakeholder in the supply chain.

[0225]

[0254] FIG. 7 is a conceptual diagram of identifying the current quality of a produce. The quality assessment system 104 can identify the current quality of a particular produce. The system 104 can also identify the current quality of the same type of produce contained in a batch of produce, e.g., a produce container. As described throughout this disclosure, the quality assessment system 104 can be configured to identify the quality of the produce when the system 104 receives a user request. That point in time can be later than when the produce is first evaluated (e.g., when the evaluation system 102 first identifies and labels the produce and identifies quality characteristics associated with the produce). The system 104 can also identify the quality of the produce earlier in the supply chain, before receiving a user request or without receiving a user request (e.g., when the produce is delivered in a produce container from a storage facility to a store). As a result, the current quality of the produce can be accurately identified early in the supply chain to make proactive upstream supply chain changes.

[0226]

[0255] To identify the current quality of the produce, the system 104 may retrieve information associated with the produce (Step A) as described throughout this disclosure. The retrieved information may include, but is not limited to, user-captured measurements 700, timestamps of user-captured produce measurements 702, produce quality models 704A-N, stored produce information 706, supply chain information 708, and quality metrics attributable to volatile substances 712. The information 700-712 may be retrieved, accessed, and / or received from one or more different computers, systems, servers, and / or devices. For example, the user-captured produce measurements 700 and timestamps 702 may be received with a user request for produce quality information from a user device. The produce quality models 704A-N and stored produce information 706 may be retrieved from one or more data stores. The supply chain information 708 may be retrieved from devices of relevant stakeholders (e.g., grocers, distribution center workers, store employees, farmers, vendors, etc.) and / or retrieved from data stores. The quality metrics 712 attributable to the volatile materials may be currently determined by the quality assessment system 104 or may be previously determined by the system 104 and then retrieved from a data store.

[0227]

[0256] The user-captured product measurements 700 may be data entered and / or scanned by a user device. The measurements 700 may represent a product. For example, the measurements 700 may be one or more image data captured by a camera or other optical sensor of the user device. The image data may capture only the product, a label or other product identifier on the product, a product with a label, and / or one or more products in an environment. The environment may include multiple products, a display shelf, product identifiers, etc. The measurements 700 may also include user-input values. For example, a user may manually enter a product identifier into the user device. The user may also review and select a product in a search interface on the user device. Once selected, the product identifier and other identifying information for the product may be the measurements 700.

[0228]

[0257] The timestamp 702 may indicate when a user scanned the product and requested the quality of the product. The timestamp 702 may be used by one or more of the models 704A-N to identify the quality of the product at the current time, which is the time indicated by the timestamp 702. Thus, one or more of the models 704A-N may be trained to correlate the time indicated by the timestamp 702 with expected quality characteristics of the product throughout the product's lifecycle.

[0229]

[0258] As described throughout, models 704A-N may be trained using machine learning techniques to identify products in measurements 700, identify quality characteristics of the products, correlate the quality characteristics with a current point in time as indicated by timestamp 702, and identify a current overall quality of the product. One or more of models 704A-N may also be trained to predict the quality of the product over the product's remaining shelf life or life cycle.

[0230]

[0259] Stored product information 706 may include historical information about the particular product or other products of the same type. Historical information may include the product's previous typical or expected ripening, bruising, spoilage, mold growth, softening, etc. Historical information may also include the product's previous pricing metrics and discounts. Product information 706 may also include storage conditions, shipping conditions, ripening conditions, historical shelf life, and production location conditions as they relate to the particular product or other products of the same type. Product information 706 may be used by one or more models 704A-N to fine-tune and more accurately identify the current quality of the product.

[0231]

[0260] Similarly, the supply chain information 708 may be used by one or more of the models 704A-N to more accurately identify the current quality of the produce. As an illustrative example, the produce may be an avocado, which is typically ripe to eat after three days. However, the supply chain information 708 may indicate that there was a delay in the transportation of this particular avocado, which delayed the transportation of the avocado to the supermarket by four days. The supply chain information 708 may also indicate that refrigeration was maintained during the delay, but no ripening agent was applied, and no process was applied to slow the ripening of the avocado. Thus, the quality assessment system 104 may typically determine that the avocado should be ripe to eat at the current time, but is no longer ripe to eat at the current time due to a delay in transportation. Thus, the supply chain information 708 may be used to more accurately identify the current quality of the produce.

[0232]

[0261] The quality metrics 712 resulting from volatiles may be used to validate the maturity curve generated by the quality assessment system 104. The metrics 712 may also be used to update and / or further train one or more models used to determine produce quality characteristics such as maturity. For further discussion of determining the metrics 712, see FIG. 1B.

[0233]

[0262] The quality assessment system 104 may process the product information (Step B). Processing the information may include applying one or more of the product quality models 704A-N to the user-captured product measurements 700. The models 704A-N may also receive the timestamp 702 and supply chain information 708 as inputs to determine the quality of the product. Additionally, the stored product information 706 may be used to further refine or otherwise determine the quality of the product at the time indicated by the timestamp 702, as described above.

[0234]

[0263] By processing the information, the quality assessment system 104 can generate a quality curve for the product (Step C). The product quality curve 710 can map the quality of a particular product over a particular period of time including the current time, as indicated by the timestamp 702. For example, the initial time represented in the product quality curve 710 can be the time of an initial assessment of a particular product by the assessment system 102.

[0235]

[0264] The product quality curve 710 may be output from one or more of the models 704A-N. One or more of the models 704A-N may be trained to generate different quality curves based on information such as the production location, different time frames during the product's life cycle, storage conditions, and other information about the product. As described above, the product quality curve 710 may also be adjusted based on information such as the stored product information 706 and supply chain information 708. The quality assessment system 104 may also generate one or more quality curves. For example, a quality curve may be generated for each quality characteristic identified by the models 704A-N. Each of the quality curves may then be correlated with each other to identify the overall quality of the product at the current time.

[0236]

[0265] In the example produce quality curve 710 of FIG. 7, the produce may be an avocado. The curve 710 shows the ripeness stages of an avocado over a seven day time frame. One or more different time frames may be used. The time frame may be based on a predicted ripeness curve or other quality curve of the produce from when the produce leaves the farm to the last time the produce may be consumed by an end consumer. Here, the curve 710 may show that on day 1, the avocado leaves the farm and is still unripe. By day 7, the quality assessment system 104 predicted that the avocado will be overripe. In other words, the quality assessment system 104 may predict that day 7 may be the last day the avocado is consumable.

[0237]

[0266] Optionally, the quality assessment system 104 may adjust the produce quality curve 710 based on the quality metric 712 attributable to volatiles (step D). For example, the models 704A-N may generate the produce quality curve 710 shown in FIG. 7 with a current ripeness stage of 3 at the current timestamp 702. However, based on analysis and modeling of the volatiles measured at the current timestamp 702, the quality assessment system 104 may determine that the actual ripeness stage of the produce is 2.5. In step D, the system 104 may then adjust the produce quality curve 710 such that the current timestamp 702 corresponds to ripeness stage 2.5 instead of 3. In doing so, the accuracy of the produce quality curve 710 may be improved. Additionally, the system 104 may use the metric 712 to further train the models 704A-N in step D to improve the accuracy of the quality curve generation and other quality metrics described herein. The metric 712 may also be used to validate the product quality curve 710 in step D (and thus validate the models 704A-N used in steps B and C).

[0238]

[0267] Using the produce quality curve 710, the quality assessment system 104 can identify the current produce quality (step E). For example, the quality assessment system 104 can determine which day on the curve 710 correlates to the timestamp 702. As shown in FIG. 7, the current point in time 702 is day 3 on the produce quality curve 710. On day 3, a particular produce, an avocado, is at ripeness stage 3. The quality assessment system 104 can also determine that the produce is approaching ripeness, which is day 4. Additionally, the quality assessment system 104 can determine that the produce has approximately four days of remaining shelf life before it is no longer consumable.

[0239]

[0268] The quality assessment system 104 may output the current quality of the product (step F). As described herein, the quality assessment system 104 may transmit the current quality back to the user device for display to the user. The quality assessment system 104 may also transmit the current quality to the pricing system 202, which may determine a price adjustment and / or discount based on the current quality. Additionally, the quality assessment system 104 may store the current quality identification. Once stored, the current quality identification may be used for subsequent quality identification of other products of the same type. The current quality identification may also be used to train and improve the assessment system 104 and the models 704A-N used for product quality identification.

[0240]

[0269] 8 is a graphical depiction of products of the same type having different quality levels at a current time. The products shown in FIG. 8 may be of the same type but from different production locations. The different production locations may affect the overall quality of the products at the same current time. Furthermore, the different production locations may affect the overall trend of the quality of each of the products of the same type as shown herein.

[0241]

[0270] As shown in this illustrative example, quality curve 802 for product location A, quality curve 804 for product location B, quality curve 806 for product location C, and quality curve 808 for product location D map the maturity over a five day period for products of the same type but from different locations A-D.

[0242]

[0271] The quality assessment system 104 generated the quality curve 802 for the product location A using the input 810. The input 810 may include a product label 812A, a model of location A 814, a product model 816, and a current timestamp 818. The product label 812A may be used to identify the particular product being modeled in the curve 802. The label 812A may also be used to retrieve additional information about the particular product being modeled in the curve 802. The model of location A 814 may be trained to identify the quality of the product based on the situation associated with location A (e.g., movement, storage, ripening, transportation). The product model 816 may be trained to identify the quality characteristics of the same type of product regardless of where the product is produced. Finally, the current timestamp 818 may represent the time when the request for the quality of the product was received by the quality assessment system 104. The quality curve 802 for the product location A indicates that the product from location A has reached maturity stage 5 at the current timestamp 818, which is day 3.

[0243]

[0272] The quality assessment system 104 used inputs 820 to generate the quality curve 804 for product location B. The inputs 820 may include product label 812B, a model 822 for location B, a product model 816, and a current timestamp 818. The product label 812B identifies a specific product that is the same as the product modeled in the curve 802, but from a different production location. Because the production locations are different, the quality assessment system 104 uses the model 822 associated with the production location: location B of the product 812B. The product model 816 may be the same as the model 816 used to generate the quality curve 802 for product location A, because the products modeled in both curves are of the same type. Additionally, the current timestamp 818 may be the same for both curves 802 and 804 to illustrate how products of the same type from different production locations may have different current qualities. Thus, the quality curve 804 for produce location B indicates that at the current timestamp 818, day 3, produce from location B has reached maturity stage 2.

[0244]

[0273] The quality assessment system 104 used inputs 824 to generate the quality curve 806 for product location C. The inputs 824 may include product label 812C, model 826 for location C, product model 816, and current timestamp 818. Product label 812C identifies a particular product that is the same as the product modeled in curves 802 and 804, but from a different production location. Because the production locations are different, the quality assessment system 104 uses model 826 associated with the production location of product 812C: location C. Product model 816 may be the same as the model 816 used to generate the quality curves 802 and 804 for product locations A and B, because the products modeled in all three curves are of the same type. Additionally, current timestamp 818 may be the same for curves 802, 804, and 806 to illustrate how products of the same type from different production locations may have different current qualities and expected qualities over time. Thus, the quality curve 806 for product location C indicates that at the current timestamp 818, day 3, the product from location C has reached maturity stage 3.

[0245]

[0274] The quality assessment system 104 used inputs 828 to generate a quality curve 808 for product location D. The inputs 828 may include a product label 812D, a model 830 for location D, a product model 816, and a current timestamp 818. The product label 812D identifies a particular product that is the same as the product modeled in the curves 802, 804, and 806, but from a different production location. Because the production location is different, the quality assessment system 104 uses a model 830 associated with the production location: location D of the product 812D. The product model 816 may be the same as the model 816 used to generate the quality curves 802, 804, and 806 for product locations A, B, and C, because the products modeled in all four curves are of the same type. Additionally, the current timestamp 818 may be the same for the curves 802, 804, and 806 to illustrate how products of the same type from different production locations may have different current qualities. Thus, the quality curve 808 for product location D indicates that at the current timestamp 818, which is day 3, the product from location D has reached a maturity stage of 2.25.

[0246]

[0275] 9A and 9B are system diagrams of components that may be used to implement the techniques described throughout this disclosure. FIG. 9A illustrates components of a computer system that may be used. FIG. 9B illustrates data stores that may be used. With reference to both FIG. 9A and FIG. 9B, the evaluation system 102, the product containers 120A-N, the quality assessment system 104, the pricing system 202, the user devices 106A-N, the product data store 118, the product model data store 116, and the pricing model data store 204 may communicate via a network 110. As described herein, one or more of the systems may be combined into a computing system, server, network, etc. Additionally, one or more of the data stores may be combined.

[0247]

[0276] The evaluation system 102 may include a product identification engine 902, a label generator 904, a quality model generator 906, and a communication interface 908. The product identification engine 902 may be configured to identify a product during an initial evaluation. For example, the engine 902 may receive image data of the product. The engine 902 may extract a portion of the image data that includes one of the products (if multiple products are present in the image data) or only a piece of the product. Using one or more machine learning and / or deep learning techniques, algorithms and / or models (e.g., convolutional neural networks), the engine 902 may identify and label features in the extracted portion of the image data that may associate the imaged product with a type of product. Additionally, the engine 902 may use object detection techniques to accurately identify the product. Any information identified or learned about the product by the product identification engine 902 may be stored in the product data store 118.

[0248]

[0277] The product data store 118 may include product profiles 944A-N. Each product identified by the evaluation system 102 may be assigned a profile. The profile may be a collection of information received or learned by the evaluation system 102 and the quality assessment system 104 and / or pricing system 202. The product profiles 944A-N may include an identifier 946, origin information 948, storage conditions 950, historical ripening information 952, historical quality information 954, and pricing information 956, respectively. Any of the information 948, 950, 952, 954, and 956 may be dynamically updated in real time as evaluations and identifications are made by the evaluation system 102, the quality assessment system 104, and the pricing system 202. Additionally, any of the information 948, 950, 952, 954, and 956 may be dynamically updated in real time based on information received or stored by relevant stakeholders throughout the supply chain.

[0249]

[0278] The label generator 904 may be configured to generate a unique identifier (e.g., barcode, label, sticker, QR code, SKU) for the product once the product is identified. The unique identifier 946 may be used to identify the particular product throughout the supply chain. The unique identifier 946 may also be used to associate quality information with the particular product. The unique identifier 946 may also be used to store and retrieve information about the particular product from one or more of the data stores 118, 116, and 204.

[0250]

[0279] The quality model generator 906 may be configured to generate one or more machine learning models that identify specific characteristics of a particular product. The models may be trained to identify characteristics indicative of the quality of the product when the quality assessment system 104 receives a request for the quality of the product, as described throughout this specification. The generator 906 may generate multiple models per product identified by the product identification engine 902. For each example, each model may be configured to identify different quality characteristics of the particular product, as described throughout this disclosure. The quality model generator 906 may also be configured to improve existing models based on subsequent quality identifications made by the quality assessment system 104. The generator 906 may also improve existing models using information received about the supply chain and / or the particular product from relevant stakeholders throughout the supply chain.

[0251]

[0280] The models generated by the quality model generator 906 may be stored in the product model data store 116. The product data store 116 may include different types of models, such as product identification models 958A-N, product quality models 960A-N, and product maturity models 964A-N. The product data store 116 may also include product model training data sets 968A-N. The product identification models 958A-N may be used by the product identification engine 902 of the evaluation system 102. One or more of these models 958A-N may also be used by the product identification engine 910 of the quality assessment system 104. The models 960A-N and 964A-N may also include product identifiers 962 and 966, respectively. The identifiers 962 and 966 may be used to associate the models 960A-N and 964A-N with a particular product.

[0252]

[0281] Additionally, as described throughout this disclosure, the quality model generator 906 can use the product model training data sets 968A-N to refine any of the models 958A-N, 960A-N, and 964A-N. The training data sets 968A-N can be associated with different products based on the product identifier 970. As an example, one training data set can be associated with avocados, while another training data set can be associated with bananas. The product identifier 970 can be used to determine which training data sets 968A-N apply to which models 958A-N, 960A-N, and 964A-N. The training data sets 968A-N can be comprised of quality assessments and identifications made by the quality assessment system 104. The training data sets 968A-N can be comprised of pricing metric adjustments and identifications made by the pricing system 202. The training data sets 968A-N can be comprised of information received from relevant stakeholders throughout the supply chain. For example, information may include supply chain status, historical ripening information, shipping status, ripening status, farm status, country of origin information, etc. Information may be continually added to the data sets 968A-N to become a robust collection of data that may be used to improve the accuracy of any of the models 958A-N, 960A-N, and 964A-N described herein.

[0253]

[0282] The product receptacles 120A-N may include sensors 122A-N, displays 136A-N, and communication interface 140. Optionally, the product receptacles 120A-N may include a power source 138, plug-in outlets 142A-N, and / or a controller 143. For further discussion of the configuration of the product receptacles 120A-N, see FIGS. 1B-1F. As described herein, the product receptacles 120A-N may be any type of structure configured to directly hold product, such as boxes, bins, containers, crates, pallets, and / or shelves. The product receptacles 120A-N may also include support structures, such as racks and / or shelves, configured to hold products or containers of products. Additionally, any of the components described with reference to the product receptacles 120A-N may be incorporated into the product receptacles 120A-N and / or may be retrofitted to existing product receptacles in a store or other parts of the supply chain lifecycle.

[0254]

[0283] In some implementations, the controller 143 can perform one or more of the techniques described herein. For example, the controller 143 can receive volatiles collected by the sensors 122A-N and process / model the volatiles using one or more product quality models 960A-N from the product model data store 116. In some implementations, the volatiles can be sent to the quality assessment system 104 for processing and analysis / modeling.

[0255]

[0284] The quality models 960A-N can be generated and trained using the disclosed techniques to identify and predict various quality metrics for various types of products. For example, the quality models 960A-N can include at least one model per product type with a sub-model for each quality metric of that product type. Each quality model 960A-N can include a respective product identifier 962, which can be used by components such as the controller 143 and / or the quality assessment system 104 to apply the appropriate model at run time. The quality models 960A-N can also include a product type A model and sub-models for product type A, which can include a product attribute A model, a product attribute B model, and a product attribute N model. The quality models 960A-N can also include a product type B model and sub-models for product type B, which include a product attribute X model, a product attribute Y model, and a product attribute Z model. Produce type A may be avocado and sub-models may include avocado ripeness, avocado firmness, avocado taste, avocado odor, etc. Produce type B may be orange and sub-models may include orange sweetness, orange sourness, orange juiciness, orange dryness, etc. As mentioned above, quality models 960A-N may include models for various other produce types as well. Quality models 960A-N may include more general models that may be applied to a variety of produce types, such as a sweetness model, a sourness model, a crispness model, and / or a juiciness model. One or more of the quality models 960A-N may also use various inputs. For example, some of the quality models 960A-N may use volatile substance measurements collected by sensors 122A-N as inputs, while other models may use produce identification scans, bar codes, or other non-invasive measurements as inputs.

[0256]

[0285] Volatiles processed and analyzed by either the controller 143 or the quality assessment system 104 may be stored as a volatile compound profile 957 in the product data 944A-N in the product data store 118. The profile 957 may include the volatiles, current quality metrics of the product, predicted quality metrics of the product and / or a quality curve determined based on the decay of the volatiles collected and analyzed for a particular product 944A-N.

[0257]

[0286] The controller 143 can determine one or more updated outputs to be presented on each display 136A-N for each product receptacle 120A-N. In some implementations, the quality assessment system 104 can determine and transmit the updated outputs to each product receptacle 120A-N for display on each display 136A-N.

[0258]

[0287] The quality assessment system 104 may include a product identification engine 910, a product quality specifier 912, a quality output generator 918, a model training engine 982, a product volatiles analyzer 984, and a communication interface 920. The product identification engine 910 may operate similarly to the product identification engine 902 of the evaluation system 102. The product identification engine 910 may receive image data or other data about a product that a user requests to be assessed by the quality assessment system 104. Using the received data and one or more of the product identification models 958A-N, the product identification engine 910 may identify the product. Identifying the product may include identifying a label or other identifier associated with the particular product.

[0259]

[0288] Once the product is identified, a product quality specifier 912 can retrieve information from one or more of the data stores that can be used to assess the current quality of the product. The product quality specifier 912 can include a data analytics engine 914 and a quality curve generator 916. The data analytics engine 914 can be configured to process information about the product that can be used to determine the current quality of the product.

[0260]

[0289] For example, the quality analysis engine 914 can apply one or more quality models to the product information. Using the product identifier identified by the product identification engine 910, the quality analysis engine 914 can retrieve one or more product quality models 960A-N and / or product maturity models 964A-N from the product model data store 116. Each of the models 960A-N and 964A-N can have a product identifier 962 and 966, respectively, that can be correlated and associated with the identifier found by the product identification engine 910. Thus, an appropriate model can be selected for quality characterization of a particular product.

[0261]

[0290] The quality curve generator 916 may be configured to generate a quality curve that may indicate the quality of the produce over a time period. The curve may indicate the quality of the produce from the time of initial evaluation by the evaluation system 102, the current time of the request for the produce quality metric, and the expected future time, as described throughout this disclosure (e.g., with reference to FIG. 7, FIG. 8). Using the generated curve, the produce quality specifier 912 may determine the current overall quality of the produce. The specifier 912 may also determine the expected future quality of the produce, the remaining shelf life of the produce, the amount of time until the produce is ready to eat, the amount of time remaining that the produce is ready to eat, etc. The produce quality specifier 912 may perform such determination using one or more machine learning trained models and / or deep learning algorithms. Furthermore, the quality determination may be stored in the produce data store 118.

[0262]

[0291] The quality output generator 918 can generate an output indicative of the current quality of the product. The output may include a visual, graphical representation, a numerical, Boolean value, and / or a string descriptor indicative of the current quality of the product as identified by the product quality specifier 912. The quality output generator 918 can also transmit the output to one or more of the user devices 106A-N (e.g., a requesting user device) for display to a user.

[0263]

[0292] The model training engine 982 may be configured to generate, train, validate, and / or update any of the models described throughout this disclosure. The engine 982 may perform similar or identical functions to the quality model generator 906 of the assessment system 102. For further discussion of model training, see FIG. 18.

[0264]

[0293] The product volatiles analyzer 984 may be configured to receive the volatiles collected by the sensors 122A-N in each product receptacle 120A-N. The analyzer 984 may process and analyze the received volatiles to identify a quality metric for each product in each product receptacle 120A-N. The analyzer 984 may include a quality identifier 986 and an output generator 988.

[0265]

[0294] The quality specifier 986 may be configured to model the volatiles using any one or more of the quality models 960A-N described above. By modeling the volatiles, the quality specifier 986 may identify current quality metrics of the product that emitted the volatiles. The quality specifier 986 may also predict future quality metrics of the product based on providing volatile data as input to the retrieved quality models 960A-N. As described with reference to FIG. 7, the quality specifier 986 may generate a product quality curve based on analysis and modeling of the volatiles emitted by the product and optionally based on one or more other non-destructive measurements of the product. In some implementations, the quality specifier 986 may retrieve a volatile compound profile 957 of the product from the product data store 118, which may then be provided as input to the retrieved quality models 960A-N.

[0266]

[0295] The output generator 988 may be configured to generate an output to be presented on each display 136A-N of each produce container 120A-N containing an assessed produce. For example, the generator 988 may generate text indicating the level of ripeness, firmness, flavor, sweetness, juiciness, or readiness of the produce in each produce container 120A-N. The generator 988 may also generate text indicating the current price, price adjustment, offer, and / or discount of the produce in each produce container 120A-N. The output may be generated based on the quality metric or other quality identification made by the identifier 986. In some implementations, the output may be generated based on identification made by other components, such as price adjustments made by the pricing system 202 and other quality identification made by the produce quality identifier 912 of the quality assessment system 104.

[0267]

[0296] The pricing system 202 may include a pricing model generator 922, a pricing engine 924, a price output generator 926, and a communication interface 928. The pricing model generator 922 may be configured to generate and train one or more models that may be used to determine price adjustments and discounts for a product based on the quality of the product. The models may be generated using training datasets that include information specific to the product and the grocer, retailer, or other relevant stakeholder in the supply chain. The training datasets may be stored in the pricing model data store 204 as pricing model training datasets 976A-N. Each of the pricing model training datasets 976A-N may also include a corresponding product identifier 978 and retailer identifier 980. The pricing model training datasets 976A-N may be updated over time to include additional information that may be used to train and refine the pricing model. For example, the pricing model training datasets 976A-N may include pricing metric identifications that have been performed by the pricing system 202. The pricing model training data sets 976A-N may also include consumer surveys that indicate what aspects of quality consumers care about when it comes to making purchasing decisions.

[0268]

[0297] As mentioned, different pricing models can be generated for different retailers. For example, one supermarket retailer may prefer a steady percentage discount on a certain type of produce every day after it reaches its prime. Another retailer may prefer to discount the price of any produce by a set amount, such as 50%, once the produce is past its prime. The pricing model generator 922 can train and generate machine learning based models according to each retailer's preferences and criteria. These models can then be associated with a particular retailer using a retailer identifier 980 in the pricing model data store 204.

[0269]

[0298] Additionally, pricing models can be generated based on quality characteristics of the produce. Each type of produce can have one or more different pricing models. As an example, one pricing model can be generated to adjust the price of avocados that are undergoing internal decay, another pricing model can be generated to adjust the price of avocados based on firmness, and another pricing model can be generated to adjust the price of avocados based on ripeness stage.

[0270]

[0299] The pricing model generator 922 can also use information received from relevant stakeholders in the supply chain to refine the generated pricing models. For example, a retailer may decide to put all ripe produce out during a holiday period. This information can be provided to the pricing model generator 922 and used to modify and train one or more models accordingly for that particular retailer.

[0271]

[0300] The pricing models generated by the pricing model generator 922 may be stored in the pricing model data store 204. The models may be stored as product pricing models 972A-N. Each of the product pricing models 972A-N may also be stored with a corresponding product identifier 974 and / or retailer identifier 975. Thus, the product pricing models 972A-N may be identified and searched for in the data store 204 using the product identifier 974 and / or retailer identifier 975.

[0272]

[0301] The pricing engine 924 may be configured to determine price adjustments and / or discounts in real-time. The pricing engine 924 may also determine the amount of time for which the adjustment or discount may be applied. Such determination may be based on retailer preferences. Additionally, price adjustments or discounts may be provided by retailers, suppliers, or other relevant stakeholders in the supply chain.

[0273]

[0302] The pricing engine 924 may be triggered when the quality assessment system 104 receives a request for quality information of the scanned product. Once the quality assessment system 104 identifies the current quality of the product, the quality information may be sent to the pricing system 202. The pricing engine 924 may apply one or more of the pricing models 972A-N to the quality information to determine a pricing adjustment for the particular product. The pricing engine 924 may include one or more modules. For example, the engine 924 may include a shelf life determiner 930, a current price determiner 932, a price reduction determiner 934, and a next price reduction determiner 936.

[0274]

[0303] The shelf-life identifier 930 may be configured to identify a remaining shelf-life of a product. The shelf-life identifier 930 may access historical information about a particular product from the product data store 118 and / or standards and procedures of the retailer that is stocking the product. The shelf-life identifier 930 may also retrieve one or more models from the model data store 116 and / or 204. Using the quality information received from the quality assessment system 104, the information from the data store 118 and the models from the data store 116 and / or 204, the shelf-life identifier 930 may identify an amount of time remaining shelf-life of the product before the product is no longer consumable or desirable for purchase. The shelf-life identifier 930 may also identify protocols for de-shelving the product, shelf placements and / or pricing decisions that correspond to the remaining shelf-life of the product. The determination made by shelf life determiner 930 may be used by one or more modules of pricing engine 924 in identifying pricing metric adjustments for the product.

[0275]

[0304] The current price determiner 932 may be configured to determine a price for a product. The price may not be tailored or specific to the current quality of a particular product. Instead, the price may be general and applicable to all products of the same type, regardless of the quality of the product. The current price determiner 932 may retrieve one or more pricing models 972A-N from the pricing model data store 204 that correspond to the particular product and retailer or other relevant stakeholder. In some implementations, the current price determiner 932 may also receive information about current market conditions and / or competitor pricing, which may be used in determining the current price of the product.

[0276]

[0305] The discount determiner 934 may be configured to determine an adjustment to the price of the produce based on the current and / or expected quality of the produce. The discount may be based on the remaining shelf life of the produce identified by the shelf life determiner 930. The discount may also be based on any quality information received from the quality assessment system 104 and / or retrieved from the produce data store 118. The discount determiner 934 may be configured to increase or decrease the current price of the produce. The discount determiner 934 may also take into account the prices or discounts of other produce of the same type to determine the pricing metric adjustment of the particular produce. For example, if the produce is an avocado and the avocado is not yet ripe for eating, the discount determiner 934 may increase the price of this particular avocado. The price increase may deter the consumer from purchasing the particular avocado. As a result, the consumer may be more inclined to purchase less expensive avocados that may already be ripe or past their ripeness. This can be advantageous in reducing or eliminating waste in the produce base. The price of a particular avocado can then be adjusted at a later time to reflect the quality of the avocado at a subsequent time (e.g., the avocado may reach ripeness and the price may begin to drop).

[0277]

[0306] The next price reduction determiner 936 may be configured to predict when the price of the produce will be adjusted in the future and the amount of the adjustment. In some implementations, the next price reduction determiner 936 may receive a predicted quality status of the produce from the quality assessment system 104. For example, the quality curve generator 916 of the quality assessment system 104 may generate a ripeness curve for a particular avocado indicating the current ripeness of the avocado and the predicted ripeness over the remaining shelf life of the avocado. The next price reduction determiner 936 may receive the ripeness curve and apply one or more machine learning trained models to determine where along the ripeness curve one or more price adjustments can be made. The next price reduction determiner 936 may also correlate those points along the ripeness curve with the number of days from the current time. For example, an avocado may be ripe at the current time. The ripeness curve may indicate that the avocado will be past ripe in two days from the current time. Thus, the next price reduction determiner 936 may predict that the next price reduction is two days from the current time. The next price reduction determiner 936 may apply one or more additional models to determine that the next price reduction is two days from the current time. In some implementations, the next price reduction determiner 936 may communicate with the discount determiner 934 and request the discount determiner 934 to identify price reductions that may occur in two days from the current time.

[0278]

[0307] The next price reduction determiner 936 can also predict future price changes based on the remaining shelf life determination of the shelf life determiner 930. In some implementations, the next price reduction determiner 936 can also predict future price changes based on information received from stakeholders throughout the supply chain. The information can include retailer preference changes, current or future retailer sales or promotions, market changes, competitor changes, etc.

[0279]

[0308] The price output generator 926 may be configured to generate an output that indicates pricing information for the produce. The output may include the current price of the produce, the current price reduction, when the next price reduction will occur, and / or what the next price reduction will be. Different outputs may also be generated based on the quality of the produce and the type of customer that requested the price information. For example, the price output generator 926 may generate a different output to a grocer at a checkout lane than the output to a consumer at the shelf. The output to the grocer may indicate an adjusted price for the produce, and the adjusted price may be automatically applied to the consumer's bill / receipt. The output to the consumer may indicate a discount that the consumer should expect if they decide to purchase the produce. In other words, the output to the consumer may not automatically apply the adjusted price to the consumer's bill / receipt.

[0280]

[0309] The price output generator 926 may also be configured to generate coupons that can be applied at the time of purchase. For example, the coupons may be digital and the consumer may store the coupons on a user device. The consumer may then present the digital coupon at the checkout counter so that the discount can be applied to the consumer's bill / receipt. The coupons may also be physical coupons that can be printed on an in-store device. The device may be placed near the produce shelves so that the consumer can obtain the corresponding physical coupon after receiving information about the quality and price of the produce on the user device. The user may then present the physical coupon at the checkout counter.

[0281]

[0310] The pricing metric identifications, adjustments, and / or discounts made by the pricing system 202 may be stored in the product data store 118 with respect to the corresponding product profiles 944A-N.

[0282]

[0311] The user devices 106A-N can be used by different types of users to request quality information about a product. Users may include, but are not limited to, supermarkets, retailers, vendors, grocers, store employees, and any other relevant stakeholders in the supply chain. The user devices 106A-N may include computers, tablets, laptops, mobile devices, smartphones, and / or cell phones. For example, a shopper at a supermarket may use a smartphone to request quality information about a product that the shopper wants to purchase.

[0283]

[0312] The user devices 106A-N may include at least a scanning device 938, a product information application 940, and a communication interface 942. The scanning device 938 may include an optical sensor that captures information about a product in the user's environment. For example, the scanning device 938 may be one or more cameras, such as a smartphone camera. The scanning device 938 may also include a barcode scanner or reader or other similar scanning device. As an example, a shopper may take a picture of a product using the scanning device 938. The shopper may take multiple pictures of the product. The shopper may also provide other input to the user devices 106A-N using one or more input devices, such as a keyboard, microphone, and / or touch screen. For example, the shopper may manually type in the name of the product or other product identifier.

[0284]

[0313] The product information application 940 may be downloaded or otherwise accessible to the user devices 106A-N. The product information application 940 may be presented to a user via a graphical user interface (GUI) display of one or more output devices (e.g., touch screen, display, etc.) of the user devices 106A-N. As described further below, a user may use the product information application 940 to request and view information about the quality of a product (see, e.g., FIGS. 13-16). The product information application 940 may also be used to view pricing information for a particular product.

[0285]

[0314] As an illustrative example, the user device may be a shopper's mobile phone. The shopper may take a picture of an avocado using a scanning device 938 (e.g., a camera) on the mobile phone. The mobile phone may send a quality information request to the quality assessment system 104. The request may include a picture of the avocado. The produce information application 940 may then present the quality information identified by the quality assessment system 104 to the shopper.

[0286]

[0315] As another example, if the user is a store employee, the produce information application 940 can present the employee with information such as ripeness trends of the produce the employee is using to restock the shelves. The produce information application 940 can also provide suggestions for where the employee should place the produce on the shelves based on the current quality of the produce.

[0287]

[0316] As yet another example, the user device may be a POS terminal at a checkout lane. A produce information application 940 at the POS terminal may present the grocer with an adjusted price for the produce scanned by the grocer at the POS terminal.

[0288]

[0317] Finally, the communication interfaces 908, 920, 928 and 942 of the evaluation system 102, the quality assessment system 104, the pricing system 202 and the user devices 106A-N may be configured to provide communication between one or more of the components, systems and data stores described herein.

[0289]

[0318] FIG. 10A is a schematic flow chart of a process 1000 for evaluating a product. One or more blocks in the process 1000 may be performed before the product is displayed in a supermarket or otherwise presented to an end consumer. For example, the process 1000 may be performed when the product arrives at a distribution center or another storage facility in the supply chain. One or more blocks in the process 1000 may be performed by the evaluation system 102, the quality assessment system 104, and / or the pricing system 202. The process 1000 may also be performed by any other computer, device, and / or computer network, device network, and / or server. For purposes of explanation, the process 1000 is described from the perspective of a computer system. The computer system may include one or more of the evaluation system 102, the quality assessment system 104, and / or the pricing system 202.

[0290]

[0319] Referring to process 1000 in FIG. 10A, at 1002, product data can be received by a computer system. The data can include image data, spectral data, x-ray data, and / or any other type of data that can be used by the computer system to identify the product. For example, a distribution center worker can take an image of the product as it arrives at the distribution center. As another example, the product can be automatically scanned by one or more imaging or scanning devices as it arrives at the distribution center. The product can move along a conveyor belt, and one or more imaging or scanning devices can be positioned along the conveyor belt to continuously capture data (e.g., images) of the product as it moves through the distribution center.

[0291]

[0320] At 1004, using the received data, the computer system can identify the product. As described throughout this disclosure, one or more object recognition techniques and machine learning trained models can be used to identify the product. For example, the computer system can receive images of multiple products. The computer system can identify each product in the images by generating a bounding box and a grid structure around each product. The computer system can then select each bounding box, extract the product therein, and perform object recognition techniques to identify specific features of the product. Using one or more machine learning trained models, the computer system can correlate the identified features to a particular type of product, thereby identifying the product.

[0292]

[0321] The computer system may also predict days until ripeness at 1006. Once the produce is identified, the computer system may apply one or more machine learning models associated with a particular type of produce to predict the ripeness of the produce over time. For example, at 1016, the computer system may apply at least one ripeness model to the produce data. As described herein, the models may be trained to identify and score produce characteristics from the image data that are indicative of ripeness stages. The models may be trained to correlate color and / or spectral data with different ripeness stages using one or more destructive and non-destructive measurements. The models may then be applied to the received produce data to non-destructively identify produce ripeness.

[0293]

[0322] The computer system may also enhance the prediction of days to eating with an assessment of volatiles of the produce emissions at 1018. As described throughout this disclosure, the volatiles emitted by the produce may be collected by sensors located around / near the produce. The volatiles may be analyzed by the computer system using one or more models to identify quality metrics of the produce. For example, the volatiles may be modeled to identify the current ripeness of the produce. The volatiles may also be modeled to predict the eating time of the produce and / or the future ripeness stage of the produce. At 1016, the output from the modeling of the volatiles may be used to confirm or adjust the prediction made by applying the ripeness model to the produce data. Thus, the prediction may be more accurate. Enhancement of the prediction with an assessment of volatiles of the produce emissions may also include further training or refining the ripeness model applied at 1016. As a result, the ripeness model may more accurately predict the days to eating of the produce represented by the produce data.

[0294]

[0323] The computer system can generate a product label at 1008. For example, the computer system can assign a unique identifier, such as a barcode, stamp, sticker, QR code, and / or SKU, to the particular identified product. The label can be used to identify the product throughout the supply chain. Distribution center workers and / or automated machines / devices can also be configured to apply / attach the label to the particular product.

[0295]

[0324] The predicted days until maturity and the produce label may then be stored by the computer system 1010. As described herein, the information may be stored in a data store, such as the produce data store 118.

[0296]

[0325] The computer system can determine 1012 a price for the identified produce based on the forecast, the produce maturity curve and / or an assessment of the volatiles of the produce emissions. For example, the computer system can determine an initial price for the produce. The initial price can be the price at which the produce is sold in a supermarket, regardless of the quality of the produce. One or more price adjustments and / or discounts can then be made to the initial price at a later time. The adjustments and / or discounts can be made if a consumer requests quality information and it is determined by the computer system that the quality of the produce has changed since the time process 1000 was performed.

[0297]

[0326] In some implementations, the initial price may be the same for all produce of the same type. The initial price may be a price set by the retailer, the store. In some implementations, the initial price may be specific to the quality or current ripeness of a particular produce. Thus, each identified produce may have a different initial price. Further, in some implementations, the initial price may be determined for one produce in the batch, and all remaining produce in the batch may receive that initial price, regardless of the quality of each remaining produce. The price may be based on the predicted number of days until ripeness. The price may also be based on one or more other quality characteristics or conditions of the produce identified by the computer system in 1004. Furthermore, the price may be based on the preferences of different stakeholders throughout the supply chain. Thus, the price may be determined based on multiple different conditions, as described throughout this disclosure.

[0298]

[0327] Still referring to 1012, the product price may be based on a product maturity curve, which may be generated by a computer system as described in FIG. 7. The curve may be generated by modeling the product data and / or volatiles collected on the product. The curve may indicate the current maturity (or other quality metric) of the product as well as the past and predicted future maturity levels of the product. The curve, especially when adjusted or verified using an assessment of the volatiles of the product emissions, may be an indication of how the price of the product may adjust over time to avoid product-based waste and encourage consumers to purchase the product before it is no longer edible or consumable.

[0299]

[0328] Similarly, the product price may also be based on an assessment of the volatiles of the product emissions at 1012. For example, the maturity level of the product may be determined from the assessment of the volatiles described throughout this disclosure. If the product has a maturity level that exceeds some threshold maturity level for that type of product, the price of the product may be decreased by a predetermined amount. On the other hand, if the product has a maturity level that is less than the threshold maturity level, the price of the product may be increased by another predetermined amount. One or more other price adjustments may be determined and made at 1012 based on the forecast, the product maturity curve, and / or the assessment of the volatiles of the product emissions.

[0300]

[0329] At 1014, the computer system can return a produce price for the produce. Returning the produce price can include outputting the produce price to a mobile application presented on the shopper's user device. Returning the produce price can include storing the price associated with the produce in a data store for later retrieval. For example, the shopper can scan the produce in a store. Based on identifying the produce from the scan, the user device can request the produce price from the data store and present the produce price to the mobile application of the user device. Returning the produce price can also include transmitting the produce price to a display on a produce container containing the produce in the store. The display can then output the price along with other information about the produce.

[0301]

[0330] In some implementations, the computer system may generate (1008) a produce label before predicting (1006) the days until eating. In some implementations, blocks 1006 and 1008 may be performed simultaneously. Additionally, the computer system may determine (1012) the produce price at the same time as predicting (1006) the days until eating. Thus, blocks 1012 and 1006 may be performed before, after, and / or simultaneously with generating the produce label at 1008 and storing the information at 1010.

[0302]

[0331] 10B is a flow chart of a process 1020 for evaluating a produce based on an assessment of volatile substances emitted by the produce. Process 1020 can be performed in combination with one or more other processes described herein, such as process 1000 of FIG. 10A, to determine supply chain changes based on the non-destructively assessed quality of the produce.

[0303]

[0332] One or more blocks in the process 1020 may be performed by the evaluation system 102, the quality assessment system 104, and / or the pricing system 202. The process 1020 may also be performed by any other computer, device, computer network, device network, and / or server, such as the controller 143 of each of the product receptacles 120A-N. For purposes of explanation, the process 1020 is described from the perspective of a computer system.

[0304]

[0333] Referring to process 1020 of Figure 10B, a machine learning model can be used to identify quality metrics and determine supply chain changes based on quality metrics from an assessment of volatile substances emitted by the product. Thus, the computer system can optionally generate a model to assess the product at 1022. The model can be generated prior to use at run time with the assessment of volatile substances. The model can also be generated during use at run time. For further discussion of training the model, see Figure 17.

[0305]

[0334] The computer system receives 1024 a signal indicative of the volatile compounds released by the produce. More specifically, the computer system can receive a signal indicative of the presence and / or amount of the released volatile compounds. The signals can be received from sensors located around / near the produce, as described with reference to FIGS. 1C-1F. The sensors can transmit one signal that aggregates all of the signals from the sensors around / near the produce. In some implementations, each sensor can transmit an individual signal to the computer system for processing.

[0306]

[0335] The computer system can receive additional information associated with the produce at 1026. The additional information can include produce type, produce age, produce storage time, produce quantity, produce storage conditions, produce transportation conditions, information about the produce production location, supply chain conditions associated with the produce, etc. The additional information can also include historical information about the produce or about the same type of produce as the produce at 1024. The historical information can include past volatile compound profiles of the produce at one or more different stages in the produce life cycle. The historical information can be used to determine, for example, whether a predicted produce quality metric based on a current volatile substance assessment matches an expected quality metric of the produce. In some implementations, the additional information can include a produce quality curve, such as a maturity curve, further described in FIG. 7.

[0307]

[0336] The computer system can generate a volatile compound profile of the product based on the received signal and product information at 1028. The profile can include information contained in the signal, such as the presence and / or amount of volatile compounds detected. The computer system can generate the volatile compound profile, which is a data structure that can be provided as input to one or more machine learning trained models described herein.

[0308]

[0337] The computer system can determine 1030 quality metrics (e.g., one or more freshness parameter values) of the produce based on the profile. For example, the computer system can provide 1032 the volatile compound profile as an input to a model associated with the produce. The model can receive and process the volatile compound profile to determine quality metrics of the produce. The computer system can then receive 1034 the quality metrics as a model output. The model can be any one or more of the geometric learning trained models described throughout this disclosure. Furthermore, the quality metrics can be any quality metric and / or freshness parameter described throughout this disclosure. The quality metrics determined can vary depending on the type and / or characteristics of the produce assessed and modeled from the volatile compounds. The quality metrics can include, but are not limited to, ripeness, freshness, shelf life, firmness, odor, sweetness, juiciness, sourness, dryness, and other quality metrics that can affect consumer decisions in purchasing and consuming the produce.

[0309]

[0338] Optionally, the computer system may adjust the product quality curve based on the quality metric at 1036. Adjusting the product quality curve may improve the accuracy of the model in non-destructively identifying the current and expected quality of the product. For further discussion, see FIG. 7.

[0310]

[0339] Additionally or alternatively, the computer system may optionally update one or more models associated with the product based on the quality metrics at 1038. The computer system may also update the models based on the adjusted quality curve of the product. As described above with reference to 1036, updating the models may increase the accuracy of the models in non-destructively identifying the current and / or predicted quality of the product.

[0311]

[0340] At 1040, the computer system can determine one or more supply chain changes for the product based on the quality metrics. Determining the supply chain changes can include determining a price adjustment (1042), identifying a shelf life (1044), and / or determining a sorting sequence (1046). See at least FIGS. 2-4 for a discussion of determining a price adjustment, at least FIGS. 1A and 4 for a discussion of identifying a shelf life for the product, and at least FIGS. 5A-5D for a discussion of determining a sorting sequence. In some implementations, the model used to identify the quality metrics at 1030-1034 can also determine supply chain changes at 1040-1046 and provide the supply chain changes as an output to the computer system.

[0312]

[0341] As an illustrative example, the computer system may determine a price adjustment by comparing the ripeness value (a quality metric) of the produce to a pricing table that includes one or more ripeness value ranges correlated with each price value, at 1042. If the ripeness value provided as an output from the produce model is within one of the ripeness value ranges, the computer system returns a price value corresponding to the price adjustment, at 1042.

[0313]

[0342] As another illustrative example, the computer system may identify the shelf life by comparing the presence and / or amount of one or more volatile compounds of the volatile compound profile to a shelf life table that includes one or more exemplary volatile compound profiles correlated with each shelf life of a particular type of produce, at 1044. If the presence and / or amount of the one or more volatile compounds is within one of the exemplary value ranges in the table, the computer system returns the corresponding shelf life to the shelf life determination, at 1044.

[0314]

[0343] As another illustrative example, the computer system includes determining 1046 a sorting sequence by comparing the quality metric to one or more sorting sequence tables that include suggested placements for particular types of produce based on the quality metric. The computer system can also compare the quality metric specified for the first group of produce to the quality metric specified for the second group of produce to determine where to place the first group of produce relative to the second group of produce. As an illustrative example, if the second group of produce is riper than the first group of produce, the second group of produce can be placed near an entrance to the store or a consumer's path of travel through the store, thereby incentivizing consumers to purchase the riper produce before the lower priced produce.

[0315]

[0344] The computer system may return the supply chain changes for the product at 1050. As described herein, the supply chain changes may be transmitted to a display of a produce container containing the product. The changes, such as a price adjustment, shelf life, or another quality metric of the product, may then be presented as an output on the display. The changes may be transmitted to a consumer's user device and may also be presented on a mobile application on the user device. The changes may be transmitted to an employee's device in the store so that the employee can place the product on the display accordingly. Returning the supply chain changes may also include storing the supply chain changes in association with the product in a data store.

[0316]

[0345] FIG. 11 is a flow chart of a process 1100 for determining a price of a product. One or more blocks in the process 1100 may be performed at the current time when a consumer or other user requests quality information for a product. For example, the process 1100 may be performed when a user scans a product on a shelf in a supermarket. One or more blocks in the process 1100 may be performed by the rating system 102, the quality assessment system 104, and / or the pricing system 202. The process 1100 may also be performed by any other computer, device, computer network, device network, and / or server. For purposes of explanation, the process 1100 is described from the perspective of a computer system. The computer system may include one or more of the rating system 102, the quality assessment system 104, and / or the pricing system 202.

[0317]

[0346] The computer system can receive produce data at 1102. As described herein, the data can be image data captured by an optical sensor of a mobile device. For example, a consumer can take a picture of a produce at a supermarket. In some implementations, the produce data can include an identification of a current quality of the produce. In some implementations, as described herein, the produce data can include a volatile compound profile of the produce. The volatile compound profile can then be analyzed by the computer system and modeled to identify produce quality metrics, such as the current quality of the produce and / or the remaining saleable shelf life of the produce.

[0318]

[0347] The computer system can identify 1104 a product in the data as described throughout this disclosure. Identifying the product can include detecting and decoding a product identifier that appears in the product data. In some implementations, identifying the product can also include determining a current quality of the product, as described herein.

[0319]

[0348] The computer system may then apply one or more pricing models to the product data at 1106. The computer system may select a pricing model associated with the identified product. For example, the computer system may use the decoded product identifier to search the data store for a pricing model that corresponds to a particular type of product. The computer system may also select a pricing model associated with a particular supermarket or other relevant stakeholder. The models may be used to determine pricing metric adjustments based on the current quality of the product.

[0320]

[0349] Quantifiable metrics of quality can be used to dynamically adjust prices. Such quantifiable metrics can be identified using one or more of the pricing models. Quality metrics can vary according to produce category. Metrics can include size, color, hardness, likelihood of mold, flavor, nutrients, pesticide use, etc. Quantifiable quality metrics can be predicted as a function of the produce's position in the supply chain by taking a limited number of upstream measurements. Thus, pricing models can be trained using large amounts of training data that can be collected using non-destructive measurement techniques. Non-destructive measurement techniques can be linked to established destructive measurements to provide accurate ground truth information on how pricing metrics can be adjusted based on quality.

[0321]

[0350] The computer system may also determine, at 1108, a remaining shelf life for the product. This determination may affect an adjustment to the price of the product. In some implementations, this determination may affect the future price of the product. This determination may be made by the computer system using one or more of the pricing models applied at 1106.

[0322]

[0351] At 1110, the computer system can determine whether the remaining saleable shelf life exceeds a threshold range. The threshold range can vary based on the type of produce and / or retailer preferences / sales. For example, some retailers may prefer to keep produce on the shelves until it is sold, regardless of the quality of the produce. Other retailers may prefer to remove produce from the shelves once the quality of the produce reaches a certain low level. Thus, the price can be reduced as the produce approaches the end of its saleable shelf life.

[0323]

[0352] If the remaining saleable shelf life exceeds the threshold range, the computer system can increase the price of the produce (1112). In other words, the produce has a long shelf life. A long shelf life can indicate that the produce is not yet ripe. A long shelf life can also indicate that it will take a longer amount of time for the produce to deteriorate. A price increase can be advantageous to discourage consumers from purchasing a particular produce until its shelf life is shorter. Thus, a higher price can attract consumers to purchase a produce with a lower price that can correspond to a shorter shelf life. If consumers purchase the lower priced produce first, waste of the produce base can be avoided or reduced. The lower priced produce can be removed from the shelves before the expiry of its saleable shelf life.

[0324]

[0353] If the remaining saleable shelf life does not exceed the threshold range, the computer system may reduce the price of the produce (1114). In other words, the produce has a shorter shelf life. A shorter shelf life may indicate that the produce is past its prime. A shorter shelf life may indicate that the quality of the produce is deteriorating. Reducing the price of such a produce may be advantageous to attract consumers to purchase this produce before the shelf life expires. As a result, produce-based waste may be avoided or reduced.

[0325]

[0354] The computer system can then return the updated price of the produce at 1116. As described herein, the updated price may be transmitted to a display on a produce container containing the produce. The updated price may be output on the display. The updated price may also be transmitted to a consumer's user device in the store. The updated price may be transmitted to a point-of-sale terminal at a checkout lane in the store. The updated price may also be stored in a data store for future retrieval and use.

[0326]

[0355] FIG. 12 is a flow chart of a process 1200 for improving a model used to determine product quality metrics. The process 1200 may be performed to continuously improve the fidelity of one or more predictive models and algorithms. One or more blocks in the process 1200 may be performed at any time during the supply chain and / or product life cycle. For example, the process 1200 may be performed after a user scans a product on a shelf and purchases the product. One or more blocks in the process 1200 may be performed by the evaluation system 102, the quality assessment system 104, and / or the pricing system 202. The process 1200 may also be performed by any other computer, device, computer network, device network, and / or server. For purposes of explanation, the process 1200 is described from the perspective of a computer system. The computer system may include one or more of the evaluation system 102, the quality assessment system 104, and / or the pricing system 202.

[0327]

[0356] Referring to process 1200, a computer system can receive product data from a user device at 1202. As described throughout this disclosure, the data can be image data. At 1204, the computer system can perform object recognition techniques to identify the environment of the product from the image data. For example, the computer system can identify the shelf on which the product is located and / or the location of the product on the shelf relative to other products or objects in the image data. In some implementations, the computer system can extract a particular color of the product identifier to normalize the lighting in the image data. Normalizing the lighting allows the computer system to more accurately identify the surrounding environment and the placement of the product in that environment.

[0328]

[0357] Once the produce is separated or otherwise extracted from the environment in the image data, at 1206, the computer system can apply one or more machine learning trained models to predict the current shelf life of the produce. In some implementations, the computer system can also predict the remaining shelf life of the produce. The current shelf life can be indicative of the current quality of the produce. For example, the current shelf life can indicate whether the produce is currently ripe, nearing or past its prime. The predicted shelf life of the produce can also be determined or otherwise adjusted / enhanced by assessment and modeling of volatile compounds emitted by the produce, as described throughout this disclosure.

[0329]

[0358] At 1208, the computer system can compare the location of the product on the shelf to the model-predicted shelf life. In other words, the environment of the product identified from 1204 can be used to identify where on the shelf the product is located. For example, the product can be located at the front of the shelf, in the middle of the shelf, or at the back of the shelf. The computer system can compare this image-based location of the product to the model-predicted shelf life to determine whether the product is located in a desirable location on the shelf. For example, products with a long remaining shelf life can be located at the back of the shelf, while products with a short remaining shelf life can be located at the front of the shelf. Placing products with a short remaining shelf life at the front of the shelf can motivate shoppers to purchase such products before products with a longer remaining shelf life.

[0330]

[0359] At 1210, the computer system can determine whether the current location on the shelf corresponds to the model-predicted shelf life. For example, the computer system can determine that if a produce is predicted to be past its prime, then the produce should be placed at the front of the shelf. Similarly, if a produce is placed at the front of the shelf, then the produce should be predicted to be past its prime. Thus, the computer system can compare the image data to the shelf life to determine how similar the supply chain is to that assumed via the model.

[0331]

[0360] If the current location on the shelf corresponds to the model predicted shelf life, the computer system can generate an output of the model predicted shelf life at 1212. In other words, the computer system can determine that the model used to predict the shelf life is accurate or at least predicts the shelf life within a predetermined margin of error. Thus, the model may not need to be refined at any particular time. Because the model is accurate, the predicted shelf life can be output to a user device and / or used by one or more computer systems to make additional predictions and identifications, such as price adjustments and discounts.

[0332]

[0361] If the current location on the shelf does not correspond to the model predicted shelf life, then in 1214 the computer system may train a model using the product data received from the user device. In other words, the computer system may determine that one or more of the models are not accurately predicting the shelf life. Thus, the models may be trained to more accurately predict the shelf life. Once the models are trained, the computer system may return to block 1206 and repeat blocks 1206-1214 until the models accurately predict the shelf life of the product.

[0333]

[0362] In some implementations, process 1200 can be used as a means to challenge a price or quality determination made by a computer system. For example, a consumer can take a picture of a product that the consumer believes was incorrectly priced. Process 1200 can be performed using the photo of the product to determine whether a price discount should be applied to the product. If the image justifies the consumer's challenge of the incorrect price, the discount can be applied.

[0334]

[0363] Figure 13 illustrates example components of a user device application interface 1300. The interface components 1300 may be presented in the product information application 940 described with reference to Figures 9A and 9B. The interface components 1300 may be presented in the user device 106A and / or any other user devices 106A-N described throughout this disclosure.

[0335]

[0364] The interface components 1300 can present information about different produce to a user. For example, a shopper can use one or more of the interface components 1300 to learn about different produce in a supermarket and decide whether to purchase a produce. The interface components 1300 can include an option 1302 to scan a produce, an option 1304 to view the highest quality produce on display, an option 1306 to view the lowest quality produce on display, an option 1308 to track total savings, an option 1310 to view environmental benefits, an option 1312 to be notified of upcoming discounts, an option 1314 to learn about the produce purchased, an option 1316 to view current discounts, an option 1318 to track the ripeness of the purchased produce, and an option 1320 to view the current ripeness of the scanned produce. One or more additional or fewer interface components 1300 can be presented, displayed, and / or selectable on the user device 106A. For example, different components may be presented, displayed, and / or selectable on a grocer user device, a store employee user device, and a shopper user device.

[0336]

[0365] The shopper may select option 1302 to scan a produce. The shopper may use an optical sensor, such as a smartphone camera, to scan the produce's produce identifier. The shopper may also manually enter a unique identifier or other information about the produce. By scanning the produce, the shopper may receive information about the produce's current quality and the price of that particular produce.

[0337]

[0366] The shopper can also select an option 1304 to view the highest quality produce on the display and an option 1306 to view the lowest quality produce on the display (see, for example, FIGS. 14A and 14B). The shopper can point their user device 106A at the entire produce display and see an augmented reality (AR) view of the highest or lowest quality produce. This can be advantageous to help the shopper decide which produce to purchase.

[0338]

[0367] A shopper can view savings over time by selecting the track total savings option 1308. Savings may include price adjustments and discounts applied to produce scanned and subsequently purchased by the shopper via the scan produce option 1302.

[0339]

[0368] By selecting the view environmental benefits option 1310, a shopper can see how purchasing a produce has helped the environment. For example, the view environmental benefits option 310 can show the amount of produce-based waste the shopper can avoid as a result of purchasing the produce. The more lower quality produce the shopper purchases, the greater the environmental benefit can be because less lower quality produce is wasted.

[0340]

[0369] The shopper can also select option 1312 to be notified of upcoming discounts. As a result, the shopper can receive alerts when a store near the shopper is discounting a particular produce. The shopper can select the type of produce for which they want to receive alerts. The shopper can also receive notifications whenever a store is discounting any type of produce. Because produce may be discounted as the current quality of the produce declines, being notified of a price reduction can motivate the shopper to purchase that produce. As a result, produce-based waste can be avoided or reduced, and the overall shelf life of the produce section can be increased.

[0341]

[0370] The shopper can select option 1314 to learn about the produce they purchased. Because each produce is assigned a unique identifier, the produce can be tracked throughout the supply chain. Thus, the movement of the produce throughout the supply chain can be presented to the shopper. For example, the shopper can see where the produce was grown, when it was shipped to the distribution center and / or store, how many hands the produce has passed through, how long the produce has been in storage, and / or when the produce was at various stages of ripeness. The shopper can also see information about who grew the produce and what production methods were used by the time the produce hit the shelves. Such transparency can be advantageous in encouraging shoppers to purchase more of the produce described throughout this disclosure.

[0342]

[0371] The shopper can select view current discounts 1316 to see which products are currently being offered at a lower price. The shopper can view current discounts on the product that the shopper scanned with the user device 106A. The shopper can also view current discounts on all products in the store. In some implementations, the shopper can also view future discounts on products.

[0343]

[0372] By selecting option 1318 to track the ripeness of the purchased produce, the shopper can continue to learn about the quality of the produce after it has left the store. Once the produce is scanned and purchased, it can be logged into the digital pantry. The digital pantry can track the time remaining before it is no longer consumable (see, e.g., FIGS. 15A and 15B). The shopper can then be notified when the produce should be eaten and / or used otherwise by the shopper. As a result, the shopper can know when to open the produce while it is still ripe to eat. The shopper can also know when to use the produce before it is wasted. In some implementations, the shopper can also learn recipes that use the produce at its current ripeness or quality stage. For example, the shopper can be presented with recipes that can be made with the produce before it is about to expire. The shopper can also be presented with suggestions about other items that can be purchased to best utilize the expiring produce. As a result, produce-based waste can be avoided even after the produce has left the store.

[0344]

[0373] A shopper can also select option 1320 to view the current ripeness of the scanned produce to learn about the quality and pricing of the scanned produce (see, for example, FIG. 16). A shopper can view the ripeness or other quality curve of the produce. A shopper can also view the current pricing, discounts, and / or future price reductions of the produce. In some implementations, a shopper can see how price adjustments, discounts, and / or initial prices were determined. This transparency in pricing can be advantageous to incentivize shoppers to use the application more and purchase more of the produce described throughout this disclosure.

[0345]

[0374] FIG. 14A is a flow chart of a process 1400 for displaying a produce based on quality using augmented reality (AR) in a user device application (see, for example, 1304-1306 in FIG. 13). FIG. 14B is an example of an AR display of a produce in a user device application. One or more blocks in the process 1400 of FIG. 14A may be performed at a current time when a consumer or other user requests quality information for a produce. For example, the process 1400 may be performed when a user scans a produce on a shelf in a supermarket. One or more blocks in the process 1400 may be performed by the quality assessment system 104. The process 1400 may also be performed by any other computer, device, computer network, device network, and / or server. For purposes of explanation, the process 1400 is described from the perspective of a computer system.

[0346]

[0375] Referring to process 1400 of FIG. 14A, a computer system can receive product image data from a user device (e.g., user devices 106A-N) at 1402. As described throughout, a shopper can take a photo of the entire shelf using the shopper's smartphone. The computer system can then identify the products in the image data at 1404. The computer system can, for example, apply one or more machine learning trained models to the image data to detect all products that may appear in the image data. Identifying all the products can include identifying a product identifier for a product and then identifying which other products in the image data are the same with that identifier. Object recognition techniques can be used to compare different products and group them based on type.

[0347]

[0376] The computer system can then retrieve 1406 produce quality information for all identified produce shown in the image data. For example, the computer system can retrieve information associated only with the identified decoded produce identifier. Although each produce may have a unique identifier, every unique identifier may nevertheless link the same type of produce to some of the same produce quality information. For example, all produce from the same production location may have the same ripeness storage status. In some implementations, the computer system can retrieve information for each individual produce based on the unique identification of that produce.

[0348]

[0377] The computer system can determine a quality of each product, at 1408. The current quality of the products can be determined using techniques described throughout this disclosure.

[0349]

[0378] Upon identifying the quality of each product, the computer system can modify an augmented reality (AR) view of the products on the shelf at 1410. In other words, the computer system can modify the presentation of the products viewed on the user device. When the user points the device at the shelf, the user can see an AR-enhanced shelf on the display of the user device. For example, the AR view can be modified such that a product identified as good quality is highlighted if that product is within the field of view of the optical sensor of the user device. Such highlighting can direct the user's attention to the product and can assist the user in selecting a product to purchase. Similarly, the AR view can be modified such that a product identified as low or not very good quality is highlighted (e.g., see FIG. 14B) if that product is within the field of view of the optical sensor of the user device.

[0350]

[0379] In some implementations, a user can click or select a highlighted product in the AR view to see more information about the product (e.g., current quality, expected quality, current price, discounts, future discounts, etc.) The information can be presented in the AR view and / or in a separate window or display screen.

[0351]

[0380] After modifying the AR view, the AR view can be transmitted to the user device at 1412, so that the user can view the entire shelf in augmented reality to assist the user in deciding which product to purchase.

[0352]

[0381] FIG. 14B illustrates an example AR view 1420 on user device 106A. The user has selected option 1304 to view the top quality produce on the display. When the user points device 106A at the shelves, an AR view 1420 of the shelves is presented to the user. As shown, the user is viewing a shelf that includes a variety of peppers, cabbages, and lettuce. Using the process 1400 shown and described in FIG. 14A, a computer system generated the AR view 1420 with the produce 1422A-N highlighted.

[0353]

[0382] Thus, it has been determined that produce 1422A-N are the highest quality produce on the display. To make this determination, the computer system may identify each type of produce by decoding produce codes 1424A-C. The computer system may also identify each type of produce by decoding a produce identifier that may be visually attached to the produce. Once each type of produce is identified, the computer system may identify which produce is within each type of produce. For example, the computer system may identify and identify orange peppers (e.g., product code 1424A). The computer system may identify and identify striped peppers (e.g., product code 1424B) and Chinese cabbage (e.g., product code 1424C). Within each type of produce (e.g., orange peppers, striped peppers, and Chinese cabbage), the computer system may identify and rank the quality of the produce. Thus, the computer system has identified produce 1422A and 1422B as the highest quality orange peppers. Produce 1422C was identified as the highest quality striped pepper. Produce 1422D and 1422N were identified as the highest quality Chinese cabbage. In some implementations, as the user moves the user device 106A in front of the shelf and different produce comes into view, the AR view 1420 can dynamically update to reflect which of the different produce is the highest quality.

[0354]

[0383] FIG. 15A is a flow chart of a process 1500 for tracking the ripeness of a purchased produce. One or more blocks in the process 1500 may be performed at any time after a user purchases a produce. For example, the process 1500 may be performed after the purchase of the produce is reflected on the user's invoice / receipt. The process 1500 may be performed at a predetermined time. The predetermined time may be in response to the remaining shelf life of the produce. For example, if the remaining shelf life of the produce is expected to be 5 days at the time of purchase, the process 1500 may be performed every day for the remaining 5 days. As another example, if the remaining shelf life of the produce is 10 days at the time of purchase, it may be preferable to perform the process 1500 every other day for the remaining 10 days.

[0355]

[0384] One or more blocks in process 1500 may be performed by the quality assessment system 104. Process 1500 may also be performed by any other computer, device, computer network, device network, and / or server. For purposes of explanation, process 1500 is described from the perspective of a computer system.

[0356]

[0385] 15A, a computer system can retrieve quality information for a purchased product at 1502. In some implementations, this can include retrieving information about the quality of the product when the product was purchased. The computer system can also retrieve historical concessions for the product and any other information that may be stored about the product. The information can be retrieved from one or more data stores, such as product data store 118 described throughout this disclosure.

[0357]

[0386] In some implementations, the computer system can also receive information from a user device. The information can include storage conditions, such as whether the user keeps the produce on the counter, in direct sunlight, in the pantry, or in the refrigerator. The information can also include input about the appearance of the produce, such as user observations of yellow or brown skin, firmness, spots, etc. The information can also include image data captured by the user after the user brings the produce home.

[0358]

[0387] At 1504, the produce quality information can be used by the computer system to determine the current quality of the produce. Techniques described throughout this disclosure can be used to determine the current quality of the purchased produce. Based on the current quality, the computer system can determine when the produce will be ready for consumption (1506). Techniques described throughout this disclosure can also be used to make such determinations. The computer system can also determine when the produce is ripe for consumption, for example, especially if the produce was purchased before it reached ripe for consumption. The computer system can also determine how much longer the produce will be ripe for consumption, for example, especially if the produce was purchased after it was ripe for consumption. The computer system can also determine the remaining shelf life of the produce before it is no longer ready for consumption.

[0359]

[0388] At 1508, the computer system can determine a recipe that uses the product. The recipe can be based on the current quality of the product. The recipe can also be based on when the product is ripe to eat. In some implementations, the recipe can be based on when the product is past its prime. Thus, the recipe can provide suggestions on how to use the product so that it is not wasted. The recipe can also suggest other products that the user can purchase to use the product.

[0360]

[0389] At 1510, the computer system can generate an output about the purchased product. The output can be presented to a user device as shown and described with reference to Figures 13 and 15B. The output can indicate the current quality of the product, the expected quality of the product, the remaining shelf life of the product, consumable readiness, and suggested recipes.

[0361]

[0390] At 1512, the output may then be transmitted to a user device for display.

[0362]

[0391] One or more blocks in process 1500 may be repeated. For example, the computer system may return to block 1504 and repeat blocks 1504-1512. Blocks 1504-1512 may be repeated at predetermined times during the remaining shelf life of the produce, at user request, and / or at any time. As a result, the quality of the purchased produce may be tracked and reported to the user at any time. The user may make better and more informed decisions about when to use the produce so that the produce can be used when it is ripe to eat and / or before the produce is no longer consumable.

[0363]

[0392] 15B is an example display for tracking the ripeness of purchased produce in a user device application. Shown in FIG. 15B is output generated by the computer system in process 1500 (see, e.g., FIG. 15A) when a user selects option 1318 for tracking the ripeness of purchased produce in the user device 106A.

[0364]

[0393] In this example, the user is purchasing two produce items: an avocado and a banana. Produce information 1550 is presented for the avocado, and produce information 1554 is presented for the banana. Produce information 1550 indicates that the user purchased an avocado with ID 101010 (e.g., a unique produce identifier, barcode, label, etc.). The avocado was purchased on March 21, 2021, and is currently "near ripe." Produce information 1550 suggests that on March 29, 2021, the avocado should be used within two days. The user can also select recipe suggestion option 1552 to see recipes that can be made with the avocado.

[0365]

[0394] Produce information 1554 indicates that the user purchased a banana having ID 121212, which was also purchased on March 21, 2021 and is currently "ready to eat." Produce information 1554 suggests that, on March 27, 2021, the banana should be eaten now. The user can also select recipe suggestion option 1556 to see recipes that can be made with the banana.

[0366]

[0395] Produce information may be presented for each produce purchased by the user. For example, if the user purchases three avocados, produce information may be presented for each of the three avocados. Finally, each avocado may ripen at a different time. One of the avocados may be currently ready to be eaten, while another avocado may no longer be ripe for consumption. Furthermore, a guacamole recipe may be presented for the avocados that are no longer ripe, but not for the avocados that are currently ready to be eaten.

[0367]

[0396] Figure 16 is an example display for viewing the current ripeness of a scanned produce in a user device application. Shown in Figure 16 is the output generated when a user selects the view current ripeness of a scanned produce option 1320 on the user device 106A. In this example, the user takes a photo of a bell pepper. The user-captured image can be presented to the user as the captured produce 1602. If the user wishes to scan a different produce or an incorrect produce is identified from the user-captured photo, the user can select the scan a different produce option 1604.

[0368]

[0397] As described throughout this disclosure, the current quality of the scanned produce can be identified and presented to the user as produce information 1606. The produce information 1606 can include the type of produce, a unique identifier associated with the produce, the current ripeness level, when the produce is expected to reach eating ripeness, the current price, when the price is expected to be discounted, and the expected discount amount of the produce. In this illustrative example, the bell pepper is identified as an orange pepper. The particular scanned produce has identifier 131313. This produce is currently "near ripe" and is expected to reach eating ripeness in 2 days on March 29, 2021. Further, the current price of orange pepper is $1.79, but on March 30, 2021, the price is expected to decrease within 3 days. The price can be reduced after the peppers reach eating ripeness, and when discounted, can motivate consumers to purchase the discounted peppers before purchasing other peppers that are not discounted. As a result, the scanned peppers may be purchased before they go to waste. As shown, produce information 1606 may also indicate that the expected markdown price on March 30, 2021 is $1.49.

[0369]

[0398] The user may also be presented with a ripeness curve 1608. The ripeness curve 1608 may show the current ripeness 1610 of the scanned produce and the ready to eat 1612 of the scanned produce. The ripeness curve 1608 may also show different stages of ripeness of the produce over time, including after the produce is placed on the shelf and scanned by the produce. In the example ripeness curve 1608, the current ripeness 1610 of the orange pepper is stage 2 on March 27, 2021. The expected ready to eat 1612 of the green pepper is stage 4 on March 29, 2021.

[0370]

[0399] To assist the user in deciding whether or not to purchase the scanned product, one or more additional or less information may be presented to the user.

[0371]

[0400] 17 is a block diagram of system components that may be used to implement a system for assessing the quality of one or more food products. Computing device 1700 is intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other suitable computers. Mobile computing device is intended to represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, and other similar computing devices. The components illustrated herein, their connections and relationships, and their functions are meant to be examples only and are not meant to limit the practice of the invention described and / or claimed herein.

[0372]

[0401] The computing device 1700 includes a processor 1702, a memory 1704, a storage device 1706, a high-speed interface 1708 connecting to the memory 1704 and a number of high-speed expansion ports 1710, and a low-speed interface 1712 connecting to a low-speed expansion port 1714 and the storage device 1706. Each of the processor 1702, the memory 1704, the storage device 1706, the high-speed interface 1708, the high-speed expansion port 1710, and the low-speed interface 1712 are interconnected using various buses and may be mounted on a common motherboard or in other suitable manners. The processor 1702 may process instructions executed within the computing device 1700, including instructions stored in the memory 1704 or the storage device 1706 to display graphical information of a GUI on an external input / output device, such as a display 1716 coupled to the high-speed interface 1708. In other implementations, multiple processors and / or multiple buses may be used as appropriate, along with multiple memories and types of memories. Multiple computing devices may also be connected, each providing a portion of the required operations (eg, as a bank of servers, a group of blade servers, or a multi-processor system).

[0373]

[0402] The memory 1704 stores information within the computing device 1700. In some implementations, the memory 1704 is one or more volatile memory units. In some implementations, the memory 1704 is one or more non-volatile memory units. The memory 1704 may also be another form of computer-readable medium, such as a magnetic disk or optical disk.

[0374]

[0403] The storage device 1706 can provide mass storage for the computing device 1700. In some implementations, the storage device 1706 can be or include a computer readable medium, such as a floppy disk device, a hard disk device, an optical disk device, a tape device, a flash memory, other similar solid state memory devices, or an array of devices including devices in a storage area network or other configuration. The computer program product can be tangibly embodied in an information carrier. The computer program product can also include instructions that, when executed, perform one or more methods, such as those described above. The computer program product can be tangibly embodied in a computer readable medium or machine readable medium, such as memory 1704, storage device 1706, or memory on the processor 1702.

[0375]

[0404] The high-speed interface 1708 manages the bandwidth-intensive operations of the computing device 1700, while the low-speed interface 1712 manages the low-bandwidth intensive operations. Such an allocation of functionality is merely an example. In some implementations, the high-speed interface 1708 is coupled to the memory 1704, the display 1716 (e.g., through a graphics processor or accelerator), and a high-speed expansion port 1710 that can accept various expansion cards (not shown). In an implementation, the low-speed interface 1712 is coupled to the storage device 1706 and the low-speed expansion port 1714. The low-speed expansion port 1714 may include various communication ports (e.g., USB, Bluetooth, Ethernet, wireless Ethernet) and may be coupled through a network adapter to one or more input / output devices, such as a keyboard, a pointing device, a scanner, or a networking device such as a switch or router.

[0376]

[0405] Computing device 1700 can be implemented in a number of different forms, as shown in the figure. For example, it can be implemented multiple times as a standard server 1720 or in a cluster of such servers. In addition, it can be implemented in a personal computer, such as a laptop computer 1722. It can also be implemented as part of a rack server system 1724. Alternatively, components from computing device 1700 can be combined with other components in a mobile device (not shown), such as mobile computing device 1750. Each such device can include one or more of computing device 1700 and mobile computing device 1750, and the entire system can be made up of multiple communication devices that communicate with each other.

[0377]

[0406] The mobile computing device 1750 includes, among other components, a processor 1752, a memory 1764, input / output devices such as a display 1754, a communication interface 1766, and a transceiver 1768. The mobile computing device 1750 may also be provided with a storage device such as a microdrive or other device to provide additional storage. Each of the processor 1752, memory 1764, display 1754, communication interface 1766, and transceiver 1768 are interconnected using various buses, and some of the components may be mounted on a common motherboard or in other suitable manners.

[0378]

[0407] The processor 1752 can execute instructions within the mobile computing device 1750, including instructions stored in the memory 1764. The processor 1752 can be implemented as a chipset of chips including separate analog and digital processors. The processor 1752 can provide coordination of other components of the mobile computing device 1750, such as, for example, the user interface, applications run by the mobile computing device 1750, and control of wireless communications by the mobile computing device 1750.

[0379]

[0408] The processor 1752 can communicate with a user through a control interface 1758 and a display interface 1756 coupled to a display 1754. The display 1754 can be, for example, a TFT (thin film transistor liquid crystal display) display or an OLED (organic light emitting diode) display or other suitable display technology. The display interface 1756 can include suitable circuitry to drive the display 1754 to present graphical and other information to a user. The control interface 1758 can receive commands from a user and convert the commands for submission to the processor 1752. Additionally, an external interface 1762 can provide communication with the processor 1752, thereby enabling short-range communication of the mobile computing device 1750 with other devices. The external interface 1762 can provide, for example, wired communication in some implementations or wireless communication in other implementations, and multiple interfaces can be used.

[0380]

[0409] The memory 1764 stores information within the mobile communication device 1750. The memory 1764 may be implemented as one or more of a computer readable medium or media, a volatile memory unit or units, or a non-volatile memory unit or units. An expansion memory 1774 may also be provided and connected to the mobile computing device 1750 through an expansion interface 1772, which may include, for example, a SIMM (single in-line memory module) card interface. The expansion memory 1774 may provide additional storage space to the mobile computing device 1750 or may store applications or other information for the mobile computing device 1750. In particular, the expansion memory 1774 may include instructions for executing or supplementing the above-mentioned processors, and may also include secure information. Thus, for example, the expansion memory 1774 may be provided to the mobile computing device 1750 as a security module and may be programmed with instructions that enable secure use of the mobile computing device 1750. Additionally, secure applications can be provided via SIMM cards along with additional information, such as placing identifying information on the SIMM card in a non-hackable manner.

[0381]

[0410] The memory may include, for example, flash memory and / or NVRAM memory (non-volatile random access memory), as described below. In some implementations, the computer program product is tangibly embodied on an information carrier. The computer program product includes instructions that, when executed, perform one or more methods, such as those described above. The computer program product may be a computer-readable or machine-readable medium, such as memory 1764, expansion memory 1774, or memory on processor 1752. In some implementations, the computer program product may be received, for example, in a signal propagated via transceiver 1768 or external interface 1762.

[0382]

[0411] The mobile computing device 1750 can communicate wirelessly through a communication interface 1766, which may include digital signal processing circuitry if necessary. The communication interface 1766 may provide for communication under various modes or protocols, such as GSM voice calls (Global System for Mobile communications), SMS (Short Message Service), EMS (Enhanced Messaging Service) or MMS messaging (Multimedia Messaging Service), CDMA (Code Division Multiple Access), TDMA (Time Division Multiple Access), PDC (Personal Digital Cellular), WCDMA (Wideband Code Division Multiple Access), CDMA2000 or GPRS (General Packet Radio Service), among others. Such communication may occur through the transceiver 1768 using radio frequencies, for example. In addition, short-range communication may occur using Bluetooth, WiFi or other such transceivers (not shown), etc. In addition, a GPS (Global Positioning System) receiver module 1770 may provide additional navigation and location related radio data to the mobile computing device 1750, which may be used as appropriate by applications running on the mobile computing device 1750.

[0383]

[0412] The mobile computing device 1750 can also communicate audibly using an audio codec 1760, which can receive spoken information from a user and convert it into usable digital information. The audio codec 1760 can also generate sounds audible to a user, such as through a speaker in a handset of the mobile computing device 1750. Such sounds can include voice telephone call sounds, recorded sounds (e.g., voice messages, music files, etc.), and can also include sounds generated by applications running on the mobile computing device 1750.

[0384]

[0413] The mobile computing device 1750 can be implemented in a number of different forms, as shown in the figure, for example, as a cellular telephone 1780. It can also be implemented as part of a smartphone 1782, personal digital assistant, or other similar mobile device.

[0385]

[0414] Various implementations of the systems and techniques described herein can be realized in digital electronic circuitry, integrated circuits, specially designed ASICs (application specific integrated circuits), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs executable and / or interpretable by a programmable system including at least one programmable processor, which can be special purpose or general purpose, coupled to exchange data and instructions with a storage system, at least one input device, and at least one output device.

[0386]

[0415] These computer programs (also known as programs, software, software applications or codes) include machine instructions for a programmable processor and may be implemented in high-level procedural and / or object-oriented programming languages ​​and / or assembly / machine languages. As used herein, the terms machine-readable medium and computer-readable medium refer to any computer program product, apparatus and / or device used to provide machine instructions and / or data to a programmable processor (e.g., magnetic disks, optical disks, memory, programmable logic devices (PLDs)), including a machine-readable medium that receives machine instructions as a machine-readable signal. The term machine-readable signal refers to any signal used to provide machine instructions and / or data to a programmable processor.

[0387]

[0416] To provide for user interaction, the systems and techniques described herein can be implemented on a computer having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user, and a keyboard and pointing device (e.g., a mouse or trackball) that allows the user to provide input to the computer. Other types of devices can be used to provide for user interaction, for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback), and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0388]

[0417] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), middleware components (e.g., as an application server), front-end components (e.g., a client computer having a graphical user interface or a web browser that allows a user to interact with an implementation of the systems and techniques described herein), or any combination of such back-end, middleware, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0389]

[0418] A computing system may include clients and servers. Clients and servers are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.

[0390]

[0419] 18 is a flow chart of a process 1800 for training a model to assess produce quality. Process 1800 can be used to train any of the models described herein. Process 1800 can be performed to receive volatile substance measurements, volatile compound profiles, non-destructive measurements, produce images, produce barcodes or other labels / identifiers as inputs to train a model to identify current quality metrics and / or predict future quality metrics of a produce. Process 1800 can also be performed to train models per product type and / or per product attribute or other quality metric.

[0391]

[0420] One or more blocks in process 1800 may be performed by evaluation system 102 and / or quality assessment system 104. Process 1800 may also be performed by any other computer, device, computer network, device network, and / or server. For purposes of explanation, process 1800 is described from the perspective of a computer system.

[0392]

[0421] Referring to process 1800 of Figure 18, at 1802, a computer system can receive training data. The training data can include product observations (1804), historical product information (1806), destructive product measurements (1808), and / or detected product volatile compounds (1810). Any combination of training data can be received to train a particular model. Additionally, different training data can be received based on the type of model being trained, the type of product the model is being trained on, and / or the type of quality characteristic the model is being trained to identify / predict.

[0393]

[0422] The produce observations in 1804 may include values ​​of characteristics of the produce that can be identified using a sensor, such as a camera or other image sensor. The produce observations may also include human-identified characteristics or properties of the produce. The produce observations may indicate one or more freshness parameters of the produce or quality characteristics, such as different ripeness stages and / or firmness of the produce. The produce observations may be made any time throughout the life cycle of the produce, such as early in the supply chain (e.g., the time from shipment of the produce from the production location to the stock shelf in the store), during storage of the produce (e.g., when the produce is stored in a warehouse or other facility for later transportation to the store), during transportation of the produce, when the produce is displayed in the store, etc.

[0394]

[0423] The historical product information in 1806 may include attributes identified for other products of the same type and stored for that product type. Attributes may include the ripeness stage or other quality characteristics of the product that have been identified, predicted or otherwise observed over time. The historical product information may also include historical pricing information, such as pricing tables that show correlations between various quality characteristics of the product and price.

[0395]

[0424] The destructive produce measurements at 1808 may include measurements obtained by destroying the produce. The measurements may be taken using one or more devices. For example, the destructive measurements may include measurements captured by piercing the outer layer of the produce, squeezing the produce, cutting the produce, and / or tasting the produce. A penetrometer and / or a hardness tester may be used to capture the destructive produce measurements. These measurements may be taken early in the supply chain, for example, before the produce is delivered to a store and displayed for purchase. In some implementations, the produce destroyed using the techniques described at 1808 may not be sold in the store. Such destructive measurements may be labeled automatically by a computer system or manually by a human user. The measurements may be labeled with various quality attributes such as ripeness stage, hardness level, taste change, etc.

[0396]

[0425] The produce volatile compounds sensed in 1810 can be volatiles or other compounds detected by sensors during any stage of the produce life cycle. The sensors can be located on, near, or around the produce when the produce is stored, transported, or displayed as described throughout this disclosure. For example, volatile compounds can be collected before the produce is transported from the farm to a distribution center, during transport to the distribution center, upon arrival at the distribution center, while stored at the distribution center, during transport from the distribution center to the end consumer retail environment, and / or upon arrival at the end retail environment. A time series of volatile compound data (1810) can be advantageous to train a model to accurately correlate the volatile compound data with different freshness parameters such as ripeness, sweetness, taste, juiciness, firmness, and other quality attributes.

[0397]

[0426] Any data received at 1802-1810 may be labeled automatically by a computer system and / or manually by a human operator. The data may be collected and then labeled with annotations indicating various types of quality attributes such as ripeness stages, different firmness levels, taste, juiciness, etc. The labeled data may then be used to train a model to identify or otherwise predict quality attributes from the received produce data during run-time.

[0398]

[0427] The computer system can then generate one or more models using the received training data at 1812. Various machine learning techniques, such as supervised machine learning, deep neural networks, principal component analysis, and / or random forest modeling, can be used to process and correlate the volatile compounds with quality attributes identified in at least one of the product observations (1804), the historical product information (1806), and / or the destructive product measurements (1808).

[0399]

[0428] At 1814, the computer system can use machine learning techniques to map the volatile compounds to produce features (e.g., quality metrics). The produce features may include any quality metrics described herein. Optionally, at 1816, the computer system can group the volatile compounds and produce features. Clustering techniques can be used for the grouping. The volatile compounds can be grouped together into classes and analyzed to identify, as an example, how classes of volatiles correlate to different ripeness stages. Some concentrations of a particular type of volatile or group of volatiles can indicate the taste, sweetness or sourness of the produce. Some concentrations of a particular type of volatile or group of volatiles can indicate the acidity level of the produce. Some concentrations of a particular type of volatile or group of volatiles can indicate the softness or hardness of the produce.

[0400]

[0429] At 1818, the computer system can then output the models for use at run-time. Outputting the models may include storing the models in a data store. In some implementations, the models can be updated and / or refined using additional data received throughout the supply chain. For example, the models can be updated using collected volatile compounds of the products during run-time use of the models. The models can also be updated using outputs generated from these models during run-time use.

[0401]

[0430] 19A-19D are conceptual diagrams of an example cabinet for assessing quality metrics of produce placed therein. More specifically, FIG. 19A shows a side view of the cabinet 1900. The cabinet 1900 can be any type of produce / food case. The cabinet 1900 can be used to ripen and / or store the produce placed therein. The cabinet 1900 can be located in a storage facility, such as a warehouse, a refrigerated storage facility, or a distribution center. The cabinet 1900 can also be located in an end-consumer retail environment, such as a store or supermarket. The cabinet 1900 can be, for example, a refrigeration unit that stores / holds produce at a predetermined temperature. In some implementations, the cabinet 1900 can be located in a transportation unit, such as a truck, a refrigerated vehicle, a boat, or other mobile storage unit. When using the cabinet 1900, volatile compounds emitted by the produce can be captured at various times / stages of the produce's supply chain to provide an accurate and efficient analysis of the produce quality.

[0402]

[0431] The cabinet 1900 can be a closed container, box, or other type of structure for holding multiple produce (e.g., a batch of produce). In some implementations, the cabinet 1900 can have an open side (e.g., the front of the cabinet 1900) for a user to access the produce therein. In some implementations, the cabinet 1900 can be closed on all sides, and produce placed therein can be accessed by opening the door of the cabinet 1900. The cabinet 1900 can hold produce of the same produce type (e.g., a pallet of all avocados). In some implementations, the cabinet 1900 can hold produce of different types (e.g., different produce having similar or different ripeness or other shelf life characteristics, different produce that are refrigerated in a retail environment such as a supermarket).

[0403]

[0432] The cabinet 1900 may have a physical structure 1901 that includes one or more shelves 1902A-N for holding / storing products. The physical structure 1901 may be a rack or other support structure for the cabinet 1900. The pallets, containers, clays, and bins or other structures described throughout this disclosure may be configured to contain groups of products on the shelves 1902A-N. In some implementations, products may be placed directly on any of the shelves 1902A-N.

[0404]

[0433] The sensors 122A-N, such as any of the sensors described throughout this disclosure, can be located on the shelves 1902A-N and / or the physical structure 1901 or its periphery within the cabinet 1900. For example, one or more of the sensors 122A-N (e.g., a group of sensors, a sensor array, etc.) can be mounted on the inside of the cabinet 1900 (e.g., the walls and / or doors of the cabinet). One or more of the sensors 122A-N can also be mounted along the face and / or length of the physical structure 1901. One or more of the sensors 122A-N can be mounted on one or more of the shelves 1902A-N. The sensors 122A-N can be configured to detect volatile compounds released by produce within the cabinet 1900. The sensors 122A-N can be located within a predetermined range from the produce, the shelves 1902A-N, and / or the container containing the produce. For example, the predetermined range can be between 1 inch and 6 inches. The predetermined range may be one or more other ranges including, but not limited to, 2 inches to 5 inches, 0 inches to 4 inches, 1 inch to 10 inches, up to 12 inches, etc. The predetermined range may vary based on what structure the sensors 122A-N are attached to (e.g., a wall of the cabinet 1900, the physical structure 1901, the shelves 1902A-N) and the distance between the structure and the produce. The sensors 122A-N in the cabinet 1900 may also be used in combination with any other sensors described throughout this disclosure, such as sensors attached to produce containers.

[0405]

[0434] The cabinet 1900 may also include valves 1905 and vacuum systems 1906 to direct airflow around the produce in the cabinet 1900 and released volatile compounds to sensors 122A-N in the cabinet (and / or in / on the produce containers). By using the valves 1905 and vacuum systems 1906, a higher concentration of released volatile compounds can be collected and used to accurately perform the techniques described herein. For example, each shelf 122A-N may have one or more openings 1903A-N on the top, bottom and / or sides of the shelf 122A-N. Hoses 1904A-N or other tubes in the hose structure 1904 may be connected to one or more openings 1903A-N of the shelves 122A-N through which air may be drawn by the vacuum system 1906 toward a sensor such as an electronic nose 1907 (e.g., e-nose, volatile substance assessment system). The hose structure 1904 may also be plumbed around the shelves 1902A-N in the cabinet 1900 (e.g., along the walls of the cabinet, the ceiling of the cabinet, the bottom of the cabinet, under the shelves of the cabinet, etc., along the top of the shelves, the bottom of the shelves, the sides of the shelves, etc.) to direct airflow from the produce to the sensors 122A-N positioned throughout the cabinet 1900. For example, the hose structure 1904 may be plumbed around the cabinet 1900 in implementations where one or more of the shelves 1902A-N do not have openings 1903A-N. In some implementations, one or more hoses, such as hose 1904B, may be located near a sensor, such as sensor 122I, to capture higher concentrations of released compounds, which may then be routed through the hose structure 1904 to the electronic nose 1907 via the vacuum system 1906 and valve 1905.

[0406]

[0435] A pump 1910 may also be attached to the cabinet 1900 (and / or part of the vacuum system 1906) and configured to draw air around the produce and into the hose structure 1904. In some implementations, the cabinet 1900 may also include one or more fans 1908A-N (which may be part of the vacuum system 1906) that help circulate air around the produce and towards one or more of the sensors 122A-N at the ends of the hose structure 1904.

[0407]

[0436] The drawn air can be routed to one or more sensors, such as a hose structure 1904, a valve 1905, and an electronic nose 1907. One or more of the sensors 122A-N, particularly the electronic nose 1907, can accumulate a collection of volatile compounds until a predetermined threshold amount of volatile compounds is collected. Once the predetermined threshold amount is met, the batch of volatile compounds can be processed by the electronic nose 1907 to generate information / data about the volatile compounds of the produce / food products. For example, the electronic nose 1907 can generate a volatile substance profile for each group of food products on the shelf 1902A-N. The volatile substance profile can include the amount of each volatile compound released by the food products and / or other data that can be used in conjunction with the disclosed techniques to identify the quality of the food products.

[0408]

[0437] Data generated by the electronic nose 1907 can be transmitted via communication interface 1914 over one or more networks (e.g., wired and / or wireless) to a computer system described herein for further processing. In some implementations, the collected data about the volatiles can be used by the processor 1912 of the cabinet 1900 to identify quality information about the food groups. Additionally, in some implementations, the cabinet 1900 can include a gas chromatograph (not shown), which can be configured in combination with or in place of the electronic nose 1907 to process and analyze the compounds collected by the sensors 122A-N.

[0409]

[0438] In some implementations, one or more of the sensors 122A-N may be configured to collect a threshold amount of emitted volatile compounds. Once the threshold amount of emitted volatile compounds has been collected, the sensors 122A-N may be heated to a predetermined temperature for a period of time (e.g., in a process called heat absorption) to release the collected volatile compounds and an accurate reading / profile of the volatile compounds may be determined (e.g., by an electronic nose 1907 and / or a gas chromatograph in combination with components of the cabinet 1900 such as the sensors 122A-N). The readings / profile may then be used by a computer system or processor 1912 to perform the techniques described herein.

[0410]

[0439] In some implementations, the sensors 122A-N may include metal oxide sensors. Infrared sensors may be used. Sometimes the sensors 122A-N may be configured in an array, where each sensor may measure a different type of volatile compound and / or a different chemical / amount of volatile compound. The cabinet 1900 described herein may also be communicatively coupled (e.g., wired, wireless) to a gas chromatograph system, which may be configured to process and analyze the volatile compounds emitted by the produce and collected / directed to the sensors through hoses in the cabinet.

[0411]

[0440] The cabinet 1900 may also include one or more display devices 1916A-N. The display devices 1916A-N may be similar or the same as any other displays described throughout this disclosure. The display devices 1916A-N may be associated with and located next to each group of products on the shelves 122A-N in the cabinet 1900. In some implementations, the display devices 1916A-N may be located next / on / near each shelf 122A-N. The display devices 1916A-N may be configured to present quality information for the groups of products in the cabinet 1900 as described throughout this disclosure. In some implementations, the cabinet 1900 may have one display device configured to present quality information for all of the products in the cabinet 1900.

[0412]

[0441] FIG. 19B shows a case 1920 for holding food items that can be placed inside a cabinet, such as the cabinet 1900 described in FIG. 19A. The case 1920 can be configured to hold groups of the same type of food items. The case 1920 can include an opening (not shown in FIG. 19B, see FIG. 19C) on the bottom of the case 1920. The hose structure 1904 described in FIG. 19A can be plumbed from the opening in the case 1920 through the valve 1905 to the electronic nose 1907 to flow / move compounds released by the food items in the case 1920 toward the electronic nose 1907. The electronic nose 1907 can be part of the case 1920, attached to the case 1920, and / or configured into the case 1920. The case 1920 and electronic nose 1907 can then be placed inside a produce cabinet, such as the cabinet 1900, a refrigeration unit, or any other type of case for holding various groups of food items. In some implementations, the electronic nose 1907 may not be part of / attached to the case 1920, but may instead be located inside a product cabinet such as cabinet 1900, in which case it may be coupled to the case 1920 via a hose structure 1904.

[0413]

[0442] The electronic nose 1907 may also communicate (e.g., wired and / or wirelessly) over a network with the display devices 1916A-N. As described herein, the display devices 1916A-N may output / present information about the food groups in the case 1920. This information may be determined based on data generated by the electronic nose 1907 about compounds released by the food groups by a computer system as described throughout this disclosure. For example, the display devices 1916A-N may output price information, discounts, ripeness stages, shelf life, taste information, and any other information that may be determined using the disclosed techniques.

[0414]

[0443] As shown in FIGS. 19A and 19B, the electronic nose 1907 can be deployed in an end consumer retail environment such as a supermarket. The electronic nose 1907 can be located under a case 1920 that holds food groups. Each hose in the hose structure 1904 can be connected to a valve 1905 (e.g., a valve switch) that can simply sample the emitted volatile compounds at different locations of the case 1920. The valve 1905 can send a sample of the compounds to the electronic nose 1907 for measurement and processing. As described above, the data is processed and displayed on the display devices 1916A-N based on the evaluation and processing of the compounds by the electronic nose 1907. In some implementations, the display devices 1916A-N can be computing devices that can control the operation of the electronic nose 1907 and / or the valve 1905. In some implementations, a remote computer system, such as any of the computer systems described herein, can control the operation of the electronic nose 1907 and / or the valve 1905. In some implementations, a processor in the produce cabinet, such as processor 1912 of cabinet 1900 in FIG. 19A, can control the operation of electronic nose 1907 and / or valve 1905. Additionally, as shown in FIG. 19B, case 1920 may not include sensors integrated with a surface of case 1920 (e.g., sidewalls, bottom, etc.). Instead, case 1920 may simply be connected to electronic nose 1907 via hose structure 1904 to collect and assess volatiles released by food placed in case 1920. In some implementations, case 1920 may include one or more sensors described throughout this disclosure, which can increase the amount of compounds collected and increase the accuracy of compound processing.

[0415]

[0444] FIG. 19C illustrates the bottom 1922 of the case 1920 of FIG. 19B. The bottom 1922 of the case 1920 can include the openings 1903A-N described with reference to FIGS. 19A and 19B. The hoses 1904A-N of the hose structure 1904 can be routed from each opening 1903A-N through the valve 1905 to the electronic device 1907 for collection and processing. The bottom 1922 can have any desired quantity of openings 1903A-N. In some implementations, the side walls of the case 1920 can also have one or more openings 1903A-N. The openings 1903A-N described herein can be sized to receive hoses 1904A-N that are approximately 1 / 16 inch to 1 / 4 inch in size. The bottom surface 1922 may have openings 1903A-N near the edges of the case 1920, along the center of the case 1920, spread across the bottom surface 1922, toward the front of the case 1920 (which may be beneficial for downwardly sloping cases in produce cabinets where food naturally follows the downward slope of the case toward the front of the case), toward the rear of the case 1920, etc. In some implementations, the more openings 1903A-N the case 1920 has, the more volatiles can be collected and assessed. The more volatiles that are collected and assessed, the more robust the data set that can be used to perform the analysis, thus improving the accuracy of further analysis of food quality.

[0416]

[0445] In some implementations, the electronic nose 1907 can include a pump to draw the collected volatile sample through the hose structure 1904, the valve 1905, and into the electronic nose 1907 for measurement. The pump can be part of the vacuum system 1906 described with reference to FIGS. 19A and 19B. Additionally, in some implementations, the electronic nose 1907 can be communicatively coupled to a pump, in which case the pump can be separate from the electronic nose 1907 and / or the case 1920. The pump can operate to control the flow of the sample from the various hoses 1904A-N through the hose structure 1904 described in FIG. 19A and into the various cases and / or shelves of food products.

[0417]

[0446] FIG. 19D shows an example case 1930 with integrated sensors 1932A-N. Case 1930 may be similar to case 1920 in that it may be configured to contain food groups. Case 1930 may be placed inside a produce cabinet described herein, such as cabinet 1900. Instead of having an opening and hose structure to deliver emitted volatile compounds to an electronic nose (see, e.g., cabinet 1920 in FIG. 19C), case 1930 includes sensors 1932A-N integrated into a surface (e.g., bottom, top, sidewalls, etc.) of case 1930. Case 1930 may include a vacuum system 1906, which may optionally include fans 1908A-N and / or pump 1910. Vacuum system 1906 may be configured to circulate air around case 1930 such that volatiles emitted by food within case 1930 are sampled by sensors 1932A-N. In some implementations, the case 1930 may be located near a separate vacuum system 1906 that may be part of a produce cabinet. In that case, the vacuum system 1906 may be configured to circulate air around the various cases, as well as the case 1930.

[0418]

[0447] The case 1930 may be integrated with a user-selected group or type of sensors 1932A-N. The sensors 1932A-N may be one or more different types of sensors that may typically be included in an electronic nose, such as electronic nose 1907. For example, any one or more of the sensors 1932A-N may be selected from a group of sensors including, but not limited to, nanocomposite sensors, electromechanical sensors, metal oxide semiconductors, photoionization sensors, and / or non-dispersive infrared sensors. As with the openings 1903A-N in the case 1920, the sensors 1932A-N may be positioned in a variety of locations on the case 1930. For more information regarding the positioning and placement of the sensors 1932A-N, see the discussion of the openings 1903A-N in FIG. 19C.

[0419]

[0448] In some implementations, the disclosed technology can be used at various other stages of the supply chain. For example, the disclosed technology can be applied in a packing facility or other storage facility. Handheld and / or stationary sensors can be located near pallets of produce. Volatiles emitted by the produce can be captured by the sensors and provided to a computer system for processing and modeling. The volatiles can be modeled to predict, for example, the ripeness stage of the produce on the pallet. Based on this prediction, the computer system can determine supply chain changes for the produce on the pallet, such as positioning the pallet in cold storage and / or a ripening room. The computer system can also utilize an inventory management system to better match the supply chain with predicted or current shelf life of the produce on a pallet-by-pallet basis.

[0420]

[0449] Although the specification contains many specific implementation details, these should not be construed as limitations on the scope of the disclosed technology or the scope of what may be claimed, but rather as descriptions of features that may be specific to certain embodiments of the disclosed technology. Certain features described herein in the context of separate embodiments may also be implemented in combination, either in part or in whole, in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented in multiple embodiments separately or in any suitable subcombination. Furthermore, features may be described herein as acting in a particular combination, and / or may initially be claimed as such, but one or more features from the claimed combination may be removed from the combination in some cases, and the claimed combination may relate to a subcombination or a variation of the subcombination. Similarly, operations may be described in a particular order, but this should not be understood as requiring that such operations be performed in a particular order or sequential order, or that all operations be performed, to achieve desired results. A particular embodiment of the invention has been described. Other embodiments are within the scope of the following claims.

Claims

1. 1. A system for identifying a quality level of a food product using non-contact assessment of said food product, comprising:

1. A quality assessment system, comprising: receiving food data, the food data including image data of a particular food; determining a current quality level of the particular food based on providing the food data as input to a food quality model, the food quality model being trained using image data of other foods having distinguishable quality characteristics, the other foods being of the same food type as the particular food; determining a number of days until the particular food product is ready to eat based on the current quality level of the particular food product; adjusting the identified number of days until ripeness based on the measured volatiles of the particular food product; determining at least one supply chain change for the particular food product based on the current quality level and the adjusted days until eating for the particular food product; A quality assessment system configured to: a display device in communication with the quality assessment system, requesting information from the quality assessment system about the quality of the particular food product, the information including at least one of the current quality level of the particular food product, the number of days until eating of the particular food product, the adjusted number of days until eating, and the at least one supply chain change; visually presenting said requested information on a display in an end consumer retail environment; and A display device configured to A system including:

2. 2. The system of claim 1, wherein the quality assessment system is further configured to generate a ripeness curve for the particular food product based at least in part on the current quality level and the measured volatiles of the particular food product, the ripeness curve indicating a current ripeness of the particular food product and a predicted ripeness level of the particular food product over a predetermined period of time.

3. The system of claim 2 , wherein the quality assessment system is configured to generate the ripening curve for a batch of food products that includes the particular food product, the batch of food products being of the same type as the particular food product.

4. The method further includes a container configured to hold a quantity of food products, including the particular food product, the container comprising: one or more sensors attached to one or more sides of the container, the one or more sensors configured to detect volatile compounds released by the food product within the container; a display screen attached to one or more sides of the container and configured to dynamically display information about the food item within the container; a communication interface configured to communicate with the quality assessment system over one or more networks; and The system of claim 1 , comprising:

5. 5. The system of claim 4, wherein the quality assessment system is configured to transmit an output indicative of the at least one supply chain change for the particular food product to the display screen of the container for presentation.

6. the at least one supply chain change is a price adjustment of the food product in the container, and the quality assessment system: determining the price adjustment for the food items in the container based at least in part on the current quality level of the particular food item; transmitting the adjusted price of the food item in the container to the display screen of the container for presentation; and The system of claim 4 configured to:

7. The quality assessment system comprises: receiving data from the one or more sensors indicative of the volatile compounds detected by the one or more sensors; providing the data indicative of the volatile compounds as input to the food quality model, the food quality model being trained to correlate volatile compounds emitted by the other foods with quality attributes of the other foods, the quality attributes being annotated in the training data of the other foods; receiving, as an output from the food quality model, the current quality level of the particular food product based on the data indicative of the volatile compounds; The system of claim 4 configured to:

8. 8. The system of claim 7, wherein determining the at least one supply chain change includes determining at least one sorting arrangement of the container relative to (i) the food product within the container and (ii) other containers that house other types of food products in the end consumer retail environment.

9. 10. The system of claim 8, wherein the sorting arrangement includes placing ripe food items closer to the front of the container and less ripe food items closer to the rear of the container.

10. 10. The system of claim 8, wherein the sorting arrangement includes placing containers having more ripe food closer to an entrance of the end consumer retail environment and placing containers having more non-ripe food further away from the entrance of the end consumer retail environment.

11. 8. The system of claim 7, wherein determining at least one supply chain change includes adjusting a price of the food item in the container based on the current quality level of the particular food item in the container.

12. 12. The system of claim 11, wherein the quality assessment system is configured to (i) increase the price of the food in the container based on a determination that the current quality level is a ripeness level below a threshold ripeness value, or (ii) decrease the price of the food in the container based on a determination that the current quality level is a ripeness level above the threshold ripeness value.

13. 8. The system of claim 7, wherein determining the one or more supply chain changes includes determining a shelf life for the food product in the container based on the current quality level.

14. The system of claim 7 , wherein the food quality model is a model that predicts ripeness of the food in the container based on the detected volatile compounds.

15. The system of claim 7 , wherein the food quality model is a model that predicts the firmness of the food in the container based on the detected volatile compounds.

16. The system of claim 7 , wherein the food quality model is a model that predicts the taste of the food in the container based on the detected volatile compounds.

17. The system of claim 7 , wherein the food quality model is a model that predicts the juiciness of the food in the container based on the detected volatile compounds.

18. The system of claim 4 , wherein the display screen is the display device.

19. The system of claim 4 , wherein the one or more sides of the container include a front wall, a rear wall, a first side wall, a second side wall, and a bottom of the container.

20. The system of claim 4 , wherein the container further comprises a local power source that is part of the container and that is configured to provide power to the one or more sensors and the display screen.

21. The evaluation calculation system further includes: receiving image data of the particular food item from an imaging device; identifying a type of the particular food product based on applying a food identification model to the image data; and identifying an initial quality level of the particular food product based on the identified type of the particular food product; generating a product identifier that uniquely identifies the particular food product, wherein information associated with the product identifier corresponds to the initial quality level of the particular food product; generating instructions to transport the particular food product to the end consumer retail environment based at least in part on the initial quality level of the particular food product; returning to at least one party in a supply chain associated with the particular food product (i) the instructions for execution by the at least one party in the supply chain, and (ii) the product identifier; The system of claim 1 configured to:

22. 22. The system of claim 21, wherein the valuation calculation system is configured to determine an initial price for the particular food product based at least in part on the initial quality level of the particular food product, the type of the particular food product, and preferences of the end consumer retail environment.

23. The system of claim 1 , wherein the particular food item is an avocado, a citrus fruit, a berry, an orange, an apple, a mango, or a vegetable.

24. The quality assessment system comprises: receiving a request for the current quality level of the particular food product from a user device in the end consumer retail environment, the request including scan data of the particular food product, the scan data including image data of the particular food product and at least one of a unique identifier affixed to the particular food product; transmitting to the user device the current quality level of the particular food item for presentation to a mobile application on the user device; The system of claim 1 configured to:

25. The system of claim 1 , wherein the display device is a consumer user device in the end consumer retail environment.

26. The system of claim 1 , wherein the display device is a visual display on a container configured to house the particular food product in the end consumer retail environment.

27. The system of claim 1 , wherein the display device is a visual display located near the particular food item within the end consumer retail environment.

28. The method further includes at least one volatile material sensor, the at least one volatile material sensor comprising: detecting volatile substances emitted by the particular food product in the end consumer retail environment; and transmitting data indicative of the detected volatile substances to the quality assessment system; The system of claim 1 configured to:

29. The quality assessment system comprises: receiving the data indicative of the detected volatile material from the at least one volatile material sensor; processing the received data to generate the measured volatiles of the particular food product; 30. The system of claim 28 configured to:

30. A container configured to hold a quantity of food, comprising: one or more sensors attached to one or more sides of the container, the one or more sensors configured to detect volatile compounds released by the food product within the container; a display screen attached to one or more sides of the container and configured to dynamically display information about the food item within the container; a communication interface configured to communicate with the quality assessment system over one or more networks; a controller in communication with the one or more sensors, the display screen, and the communication interface, the controller being configured to update information displayed on the display screen based on quality metrics of the food in the container; A container comprising:

31. 31. The container of claim 30, further comprising a local power source configured to provide power to the one or more sensors, the display screen, the controller, and the communication interface.

32. 31. The container of claim 30, further comprising at least one sensor located a threshold distance away from the container and configured to detect the volatile compounds emitted by the food product within the container.

33. 33. The container of claim 32, wherein the threshold distance is within the range of 1 to 6 inches.

34. The quality metrics of the food product in the container are determined by a computer system in communication with the container, the computer system comprising: receiving data from the controller indicative of the volatile compounds detected by the one or more sensors; providing the data indicative of the volatile compounds as input to a food quality model, the food quality model being trained to correlate volatile compounds emitted by other foods of the same food type as the food in the container with quality attributes of the other foods; receiving the quality metric of the food in the container as an output from the food quality model; transmitting, via the communication interface, to the controller an output to be presented on the display screen, the output being indicative of the quality metric of the food product; 31. The container of claim 30 configured to perform operations including:

35. 35. The container of claim 34, wherein the output indicative of the quality metric of the food product includes a current ripeness level of the food product.

36. 35. The container of claim 34, wherein the output indicative of the quality metric of the food product comprises a current firmness of the food product.

37. 35. The container of claim 34, wherein the output indicative of the quality metric of the food product comprises a current readiness of the food product.

38. 35. The container of claim 34, wherein the output indicative of the quality metric of the food product comprises a current taste of the food product.

39. 35. The container of claim 34, wherein the output indicative of the quality metric of the food product comprises a current sweetness of the food product.

40. The operation includes: determining at least one supply chain change based on the quality metrics of the food product in the container; transmitting the at least one supply chain change to the controller via the communication interface for execution; 35. The container of claim 34, further comprising:

41. The at least one supply chain change is adjusting a price of the food in the container based on the quality metric; and 41. The system of claim 40, wherein the controller is configured to (i) generate an updated output including the adjusted price of the food item in the container, and (ii) instruct the display screen to present the updated output.

42. 35. The system of claim 34, wherein the food quality model is trained to model the data indicative of the volatile compounds to a quality curve of the food in the container.

43. The food quality model is generated by the computer system: receiving training data for the other foods, the training data including (i) food observations, (ii) food history information, (iii) destructive measurements of the other foods, and (iv) detected volatile compounds from the other foods, at least one of (i)-(iv) being labeled and annotated with food features; applying machine learning techniques to the training data to generate the food quality model based on mapping the detected volatile compounds from the other foods to the food features; outputting said food quality model for use at run time; 35. The system of claim 34, which has been trained using a process comprising:

44. A case configured to hold a quantity of food, comprising: a physical structure having one or more levels for storing food groups; one or more sensors attached to the case, the one or more sensors configured to detect volatile compounds emitted by the food group at the one or more levels of the physical structure; one or more display devices configured to dynamically display information about the food groups in the case; a hose structure positioned around the food group; a vacuum system configured to draw air from around the food group through the hose structure to direct the concentration of the volatile compounds emitted by the food group towards at least one of the one or more sensors; a controller in communication with the one or more sensors, the one or more display devices, and the vacuum system, the controller being configured to update information displayed on the one or more display devices based on quality metrics of the food groups in the case, the quality metrics of the food groups being determined based at least in part on the volatile compounds emitted by the food groups; Including cases.

45. 45. The case of claim 44, wherein the physical structure is a rack.

46. 45. The case of claim 44, wherein the one or more levels are shelves.

47. 45. The case of claim 44 which is a refrigeration unit.

48. 45. The case of claim 44 which is a cabinet.

49. 45. The case of claim 44, which is a food ripening cabinet.

50. 45. The case of claim 44 which is a food ripening enclosure.

51. 45. The case of claim 44, wherein one or more sensors are mounted on one or more sides of the case proximate to the food items.

52. 45. The case of claim 44, wherein one or more sensors are attached to the physical structure in proximity to the food group.

53. 45. The case of claim 44, wherein one or more sensors are mounted on at least one of the one or more levels proximate the food group.

54. 45. The case of claim 44, further comprising one or more containers located at said one or more levels, each of said one or more containers configured to contain a respective one of said food groups.

55. 1. A system for analyzing the quality of a food product using non-contact assessment of said food product, comprising: a case configured to hold a food group; a volatile assessment system configured to collect volatile compounds emitted by the food groups within the case and to process the collected volatile compounds to generate volatile data for the food groups; a tube structure located around the case and in fluid communication with the volatile substance assessment system, the tube structure being configured to deliver the volatile compounds emitted by the food groups within the case to the volatile substance assessment system; a controller configured to control operation of the volatile substance assessment system; a communications interface configured to communicate over one or more networks with a quality assessment system, the quality assessment system configured to identify one or more quality metrics for the food group based at least in part on the volatile substance data; A system including:

56. 56. The system of claim 55, further comprising a vacuum system in communication with the volatile substance assessment system, the vacuum system configured to draw the volatile compounds released by the food groups within the case through the tubing structure and toward the volatile substance assessment system for collection.

57. 57. The system of claim 56, wherein the vacuum system includes a pump configured to draw the volatile compounds through the tubing and toward the volatile material assessment system for collection.

58. 57. The system of claim 56, wherein the vacuum system includes a fan configured to draw the volatile compounds through the tube structure and towards the volatile material assessment system for collection.

59. further comprising a valve in fluid communication with the tubing and the volatile material assessment system; 56. The system of claim 55, wherein the controller is configured to open the valve to allow the volatile compounds to pass from the tubing structure into the volatile substance assessment system for collection.

60. 60. The system of claim 59, wherein the controller is configured to close the valve based on a determination that a predetermined threshold amount of volatile compounds has been collected by the volatile material assessment system.

61. 61. The system of claim 60, wherein the controller is configured to generate instructions that, when executed, cause the volatile material assessment system to process the collected volatile compounds to generate the volatile material data.

62. 60. The system of claim 59, wherein the valve is a switch.

63. 56. The system of claim 55, further comprising a display device in communication with the volatile substance assessment system across the one or more networks, the display device configured to present information about the food groups on a display screen.

64. 64. The system of claim 63, wherein the display device is located adjacent the case.

65. 64. The system of claim 63, wherein the display device is attached to the case.

66. 64. The system of claim 63, wherein the display device is in communication with the quality assessment system over the one or more networks, and the display device is configured to present the one or more quality metrics of the food groups in the case.

67. 56. The system of claim 55, wherein the case includes one or more openings in fluid communication with one or more tubes of the tubular structure.

68. 68. The system of claim 67, wherein the one or more openings are integrated into a bottom surface of the case.

69. 68. The system of claim 67, wherein the one or more tubes are approximately in the range of 1 / 16 inch to 1 / 4 inch in size.

70. 68. The system of claim 67, further comprising a vacuum system configured to draw air through the one or more tubes of the tube structure to deliver the volatile compounds emitted by the food groups through the one or more openings in the case, through the one or more tubes and to the volatile substance assessment system for collection.

71. 56. The system of claim 55, wherein the volatile substance assessment system is an electronic nose.

72. 56. The system of claim 55, wherein the volatile substance assessment system includes one or more sensors for detecting volatile compounds emitted by the food group.

73. 73. The system of claim 72, wherein each of the one or more sensors is configured to detect a different type of volatile compound in the volatile compounds emitted by the food group.

74. 56. The system of claim 55, wherein the volatile substance assessment system is a gas chromatography machine.

75. 1. An apparatus for analyzing the quality of a food product using non-contact assessment of said food product, comprising:

1. A case configured to hold a food group, comprising: one or more sensors configured to collect volatile compounds emitted by the food items within the case; and a vacuum system configured to draw the volatile compounds emitted by the food items within the case towards the one or more sensors; a controller configured to control operation of the vacuum system; An apparatus comprising:

76. 76. The apparatus of claim 75, wherein the vacuum system includes a pump.

77. 76. The apparatus of claim 75, wherein the vacuum system includes a fan.

78. 76. The apparatus of claim 75, wherein the one or more sensors are selected from a group of sensors including at least one of a nanocomposite sensor, an electromechanical sensor, a metal oxide semiconductor, a photoionization sensor, and a non-dispersive infrared sensor.

79. 76. The apparatus of claim 75, further comprising a display device in communication with the controller over one or more networks, the controller configured to generate instructions that, when executed, cause the display device to present information about the food groups in the case, the information based, at least in part, on the volatile compounds collected by the one or more sensors.

80. 76. The apparatus of claim 75, wherein the controller is configured to generate volatile substance data for the food groups in the case based on the volatile compounds collected by the one or more sensors.

81. 81. The apparatus of claim 80, wherein the controller is configured to transmit the volatile substance data to a quality assessment system configured to identify one or more quality metrics for the food group based at least in part on the volatile substance data.

82. 81. The apparatus of claim 80, wherein the controller is configured to identify one or more quality metrics for the food group based at least in part on the volatile substance data.

83. 1. A method for dynamically identifying a quality level of a food product using non-contact assessment of said food product, comprising: receiving, by a computing system, food data, the food data including image data of a particular food; determining, by the computing system, a current quality level of the particular food product based on providing the food data as input to a food quality model, the food quality model being trained using image data of other foods having distinguishable quality characteristics, the other foods being of the same food type as the particular food product; determining, by said computing system and based on said current quality level of said particular food product, a number of days until said particular food product is ready to eat; adjusting, by the computing system, the determined number of days until eating based on the measured volatiles of the particular food product; determining, by the computing system, at least one supply chain change for the particular food product based on the current quality level and the adjusted days until eating for the particular food product; transmitting, by the computing system, to a display device, information about the quality of the particular food product to be visually presented on a display in an end consumer retail environment, the information including at least one of the current quality level of the particular food product, the number of days until eating of the particular food product, the adjusted number of days until eating, and the at least one supply chain change; The method includes:

Images