Multi-dimension-based sales prediction model method and system

Abstract

The invention discloses a sales prediction model method and system based on multiple dimensions. The method comprises the following steps: photographing a picture of a user shopping in a supermarket by using a handcart; identifying commodities put into the trolley by the user according to the shot pictures, and obtaining a list of the commodities put into the trolley by the user; obtaining a settlement order of the shopping of the user; comparing the commodity information of the settlement order with a list of commodities put into the trolley by the user to obtain commodities taken out from the trolley by the user; and according to the commodities taken out by the user, identifying whether the same category of commodities exist in the settlement order, if so, obtaining the same category of commodities in the settlement order, taking the same category of commodities as competitive commodities of the commodities taken out by the user, obtaining the selling prices of the competitive commodities from the commodity information of the settlement order, and obtaining the corresponding selling prices according to the commodities taken out by the user. And the supplier can more flexibly judge the sales condition of future commodities. And supermarkets or suppliers can predict future sales trends more accurately.

Description

Technical Field

[0001] The present application relates to the field of data processing, and specifically to a sales forecasting model method and system based on multiple dimensions. Background Art

[0002] It is crucial for supermarkets and suppliers to know which products are more popular and how they are selling. Usually, supermarkets and suppliers can accurately track the sales data of specific products, which are generated based on actual sales orders.

[0003] Despite having specific sales figures, supermarkets and suppliers often lack an understanding of some key information about the actual shopping process of customers. One important piece of information is what the main competitors of certain products are. When customers are selecting products, they often compare similar products of different brands or different types and ultimately decide to buy one or several of them, which cannot be learned by existing technologies. This limits the sales forecasts of supermarkets and suppliers. Summary of the invention

[0004] In view of the above problems, the present application provides a sales forecasting model method and system based on multiple dimensions, which solves the problem of how to find corresponding competing products for the supplier's products and provide data support for sales forecasting.

[0005] To achieve the above purpose, the inventor provides a sales forecasting model method based on multiple dimensions, comprising the following steps: Take a photo of a user shopping in a supermarket using a shopping cart; Identify the items that the user has put in the cart based on the photographed photos, and obtain a list of the items that the user has put in the cart; Get the settlement order of the user's shopping; Compare the product information of the settlement order with the list of products that the user has put in the cart, and obtain the products that the user has taken out of the cart; Identify whether there are products of the same category in the settlement order based on the product taken out by the user. If so, obtain the products of the same category in the settlement order and use them as competing products of the product taken out by the user, obtain the selling price of the competing products from the product information of the settlement order, and obtain the corresponding selling price based on the product taken out by the user.

[0006] Further, the user corresponding to the trolley includes a first user and a second user, and the first user is a child; When the items placed in the cart by the user are identified according to the photographed photos and a list of items placed in the cart by the user is obtained, the following steps are also included: It is determined through the taken photos whether the first user takes the product from the display area of ​​the shelf and puts it into the trolley. If so, it is determined whether the second user puts the product back to the display area alone or with the help of the first user within a preset time. If so, the products placed in the trolley by the first user are removed from all the products placed in the trolley to obtain a list of the products placed in the trolley by the users.

[0007] Further, the first user and the second user are identified by the following steps: Get the pixel height of the marker in the captured photo; Get the pixel height of the user in the captured photo; Get the actual height of the logo, and get the actual height of the user based on the pixel height of the logo, the pixel height of the user and the actual height of the logo; It is determined whether the actual height of the user is less than the height threshold. If so, the user is determined to be the first user; if not, the user is determined to be the second user.

[0008] Furthermore, the identification is a product label on the shelf, an end of a shelf layer, or an identification plate fixed on the layer or the ground.

[0009] Furthermore, when identifying the commodities placed in the cart by the user according to the photographed photos, the following steps are also included: Identify the products that the user has taken from the shelf placement area based on a number of photos taken before and after; Identify the product labels corresponding to the display area based on the photos taken, and obtain the weight of the product based on the product labels; Obtain the weight difference detected by the pressure sensor on the placement area; Determine whether the weight obtained according to the product label and the weight difference are less than the weight threshold. If so, it is determined that the product taken out is correctly identified. If not, use OCR technology or vision and deep learning technology to identify the correct name of the product taken out by the user from the shelf placement area and output it.

[0010] Furthermore, when identifying the commodities placed in the cart by the user according to the photographed photos, the following steps are also included: Identify products in photos using OCR technology or vision and deep learning technology.

[0011] Furthermore, the method further comprises the following steps: The competing products and their selling prices, as well as the products picked up by the user and their selling prices are sent to the suppliers corresponding to the products picked up by the user.

[0012] Furthermore, the method further comprises the following steps: Count the number of times the same product is selected by users within a preset period and the corresponding competing products; It is determined whether the frequency of occurrence of the competing product corresponding to the product selected by the user is greater than the frequency threshold. If so, a price comparison curve between the selected product and its high-frequency competing products is sent to the supplier of the product selected by the user.

[0013] Furthermore, the method further comprises the following steps: Get the past purchase date and sales amount of the items that the user has taken out from the cart; According to the sales data of the commodity, a plurality of preset date ranges and the preset activity intensity corresponding to each preset date range, a time series training data set is obtained, wherein the time series training data set includes features of the commodity and sales amounts corresponding to the features, wherein the features include identification features, date features, activity features and time series features; Performing model training using a loss function based on the time series training data set to obtain a sales amount prediction model; The sales amount prediction model is used to predict the sales amount range of the items that the user takes out from the cart.

[0014] To achieve the above objectives, the inventors also provide a multi-dimensional sales forecasting model system that stores computer program instructions, which, when executed by a processor, implement the multi-dimensional sales forecasting model method described in any of the above embodiments.

[0015] Different from the prior art, the above technical solution has the following beneficial effects: Competitive products and their prices, as well as products picked up by users and their prices, can be sent to the corresponding suppliers of the products picked up by users. Suppliers can more flexibly judge the future sales of products and adjust production plans and inventory management more flexibly. It can also allow supermarkets or suppliers to more accurately predict future sales trends.

[0016] The above-mentioned records related to the invention content are only an overview of the technical solution of the present application. In order to enable ordinary technicians in the field to more clearly understand the technical solution of the present application, and then implement it according to the text of the specification and the contents recorded in the drawings, and to make the above-mentioned purpose and other purposes, features and advantages of the present application easier to understand, the following is an explanation in combination with the specific implementation mode and drawings of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings are only used to illustrate the principles, implementation methods, applications, characteristics and effects of the specific embodiments of the present invention and other related contents, and shall not be considered as limiting the present application.

[0018] In the drawings of the specification: Figure 1 This is a flow chart of the sales forecasting model method in this embodiment; Figure 2This is a flowchart of identifying the commodities placed in the cart by the user according to the photographed photos in this embodiment, and obtaining the commodity list placed in the cart by the user; Figure 3 This is a flow chart for identifying a first user and a second user in this embodiment; Figure 4 This is a flow chart of identifying commodities placed in a shopping cart by a user based on photographed photos in this embodiment; Figure 5 This is a flow chart for counting the number of times the same product is taken out by users within a preset period and the corresponding competing products in this embodiment. DETAILED DESCRIPTION

[0019] In order to explain in detail the possible application scenarios, technical principles, specific schemes that can be implemented, and the purposes and effects that can be achieved, the following is a detailed description of the specific embodiments listed in conjunction with the accompanying drawings. The embodiments described herein are only used to more clearly illustrate the technical solutions of the present application, and are therefore only used as examples, and cannot be used to limit the scope of protection of the present application.

[0020] Reference to "embodiment" herein means that the specific features, structures or characteristics described in conjunction with the embodiment may be included in at least one embodiment of the present application. The term "embodiment" appearing in various places in the specification does not necessarily refer to the same embodiment, nor does it particularly limit its independence or association with other embodiments. In principle, in the present application, as long as there is no technical contradiction or conflict, the various technical features mentioned in the embodiments can be combined in any way to form a corresponding implementable technical solution.

[0021] Unless otherwise defined, the technical terms used in this document have the same meanings as those generally understood by those skilled in the art to which this application belongs; the use of relevant terms in this document is only for describing specific embodiments and is not intended to limit this application.

[0022] In the description of this application, the term "and / or" is an expression used to describe the logical relationship between objects, indicating that three relationships may exist, for example, A and / or B, which means: A exists, B exists, and A and B exist at the same time. In addition, the character " / " in this article generally indicates that the objects before and after are in an "or" logical relationship.

[0023] In the present application, terms such as “first” and “second” are merely used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship of quantity, priority or sequence between these entities or operations.

[0024] Without further limitations, in this application, the words "include", "comprises", "has" or other similar open-ended expressions used in sentences are intended to cover non-exclusive inclusion. These expressions do not exclude the presence of additional elements in the process, method or product including the elements, so that the process, method or product including a series of elements may include not only those limited elements, but also other elements not explicitly listed, or also include elements inherent to such process, method or product.

[0025] Similar to the understanding in the Examination Guidelines, in this application, expressions such as "greater than", "less than", "exceed" and the like are understood to exclude the number itself; expressions such as "above", "below", "within" and the like are understood to include the number itself. In addition, in the description of the embodiments of this application, "multiple" means more than two (including two), and similar expressions related to "multiple" are also understood in this way, such as "multiple groups", "multiple times", etc., unless otherwise clearly and specifically limited.

[0026] In the description of the embodiments of the present application, space-related expressions used, such as "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "vertical", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc., indicate the orientation or position relationship based on the orientation or position relationship shown in the specific embodiments or drawings, and are only for the convenience of describing the specific embodiments of the present application or facilitating the reader's understanding, and do not indicate or imply that the referred device or component must have a specific position, a specific orientation, or be constructed or operated in a specific orientation, and therefore cannot be understood as a limitation on the embodiments of the present application.

[0027] The processor described in the embodiments of the present application can be implemented by hardware, firmware, software or a combination thereof, and can use a circuit, a single or multiple application-specific integrated circuits (Application Specific Integrated Circuit, ASIC), a digital signal processor (Digital Signal Processor, DSP), a digital signal processing device (Digital Signal Processing Device, DSPD), a programmable logic device (Programmable Logic Device, PLD), a field programmable gate array (Field Programmable Gate Array, FPGA), a central processing unit (Central Processing Unit, CPU), a controller, a microcontroller, at least one of a microprocessor, and also includes other physical, biological or chemical structures that can achieve similar or equivalent functions to the processors listed above, such as biological neurons, quantum computing units, DNA computing units, etc., so that the processor can execute some or all of the steps in the computer program or method involved in the various embodiments of the present application, or any combination of the steps mentioned therein.

[0028] The computer program involved in the embodiment can be stored in a computer device readable storage medium, which includes but is not limited to a disk, a tape, a magnetic card, a floppy disk, a flash memory, an optical disk, an optical card, a read-only memory (ROM), a random access memory (RAM), an erasable programmable ROM (EPROM) and an electrically erasable programmable ROM (EEPROM), etc., and also includes other biological, physical or chemical structures that can achieve the same or equivalent functions as the storage media listed above, such as DNA, RNA, protein and other units with information storage capabilities. In a specific embodiment, the storage medium involved can be one of the above-mentioned media types, or a combination of the above-mentioned media types. In different embodiments, the computer program involved in the embodiment can be stored in a single medium in a centralized manner, or it can be stored in multiple media in a distributed manner. The memory containing the computer device readable storage medium can be a non-volatile memory or a random access memory. These computer device readable storage media can be built into the device, or they can be connected to the device involved in the embodiment as an external device or a part of an external device. In some embodiments, a memory with a computer device readable storage medium is deployed locally; in other embodiments, a solution of deploying the memory away from the processor may also be adopted, such as a network attached memory accessed via an RF circuit or an external port and a communication network, wherein the communication network may be the Internet, one or more intranets, a local area network (LAN), a wide area network (WLAN), a storage area network (SAN), etc., or a suitable combination thereof, as long as the computer device can access the memory. In addition, the computer program involved in the embodiment may be stored in plaintext / ciphertext form, or may be designed as training data, which may be integrated and reorganized through model training and implicitly stored in the parameter state of a deep neural network or other machine learning model.

[0029] See also Figure 1 This embodiment provides a sales forecasting model method based on multiple dimensions, comprising the following steps: Step S101, taking photos of a user shopping in a supermarket using a shopping cart, including photos of picking up and placing goods and photos of the shopping cart; Step S102, identifying the commodities placed in the cart by the user according to the photographed photos, and obtaining a list of commodities placed in the cart by the user; Step S103, obtaining the settlement order of the user's current purchase; Step S104, comparing the commodity information of the settlement order with the commodity list put into the cart by the user, and obtaining the commodity taken out from the cart by the user; Step S105, based on the product taken out by the user, identify whether there is a product of the same category in the settlement order. If so, proceed to step S106, obtain the product of the same category in the settlement order and use it as a competing product of the product taken out by the user, and obtain the selling price of the competing product from the product information of the settlement order, and obtain the corresponding selling price based on the product taken out by the user. If not, it means that the user has not taken out the product from the cart.

[0030] It should be noted that the main function of the camera is to capture the status of the trolley used by the user during the shopping process and obtain information such as the type and quantity of the goods. In order to ensure that the captured product information is accurate, the camera needs to have high resolution and clarity. Generally, a high-definition camera with high resolution and clarity can be selected, and a smart camera with graphics processing and analysis capabilities can also be selected as needed. According to the actual layout of the supermarket, the position of the camera is intelligently planned to ensure that each key area has sufficient coverage while avoiding excessive overlap and waste of resources. For example, multiple cameras can be installed above the supermarket shelves so that the shelves can be overlooked from the top to capture the movement path of the trolley and the goods placed in it. This can effectively avoid occlusion problems and improve recognition accuracy. Installing cameras at both ends of the aisle can not only detect the goods on the shelf at the middle height, but also monitor the situation of the trolley entering and leaving the aisle. Installing cameras near the checkout counter is not only used to record the settlement process, but also to verify whether the final purchased goods are consistent with the goods in the trolley previously photographed. The settlement order can be obtained from the sales module of the supermarket. The settlement order usually contains product data, including but not limited to product name, price, weight and other information.

[0031] The product information of the settlement order includes but is not limited to product name, selling price, weight and other information. Compare the product list in the settlement order with the product list obtained through image recognition to identify which products have been removed from the cart by the user. For those products that have not been purchased, check whether there are similar competing products in the settlement order. If so, obtain the selling price of the competing products. Of course, other data can also be obtained according to actual needs to provide data support for subsequent supermarket or supplier analysis. Help supermarkets and suppliers better understand market dynamics and consumer purchasing behavior. For example, by analyzing the price differences between different competing products and their impact on sales, pricing strategies can be optimized.

[0032] "Same category" refers to a class of products with similar functions or the same purpose. For example, different brands of toothpaste belong to the same category, and different flavors of yogurt belong to the same category. Suppose a user puts the following items into the cart while shopping in a supermarket: anti-sensitivity toothpaste from brand A, regular yogurt from brand B, and chocolate biscuits from brand C. However, at checkout, the user's final settlement order shows that they purchased the following items: anti-sensitivity toothpaste from brand A, low-fat yogurt from brand E, and oatmeal biscuits from brand F. Brand B's regular yogurt is taken out, and the user chooses brand E's low-fat yogurt. Although the specific product attributes of the two are different, they still belong to the same category (yogurt), that is, brand E's low-fat yogurt is a competitor of brand B's regular yogurt. Brand C's chocolate biscuits are taken out, and the user chooses brand F's oatmeal biscuits. Although the two are different in ingredients and types, they can both be broadly classified as snacks or biscuits. Therefore, brand F's oatmeal biscuits can be analyzed as a competitor of brand C's chocolate biscuits.

[0033] The above technical solution has the following beneficial effects: The sales forecasting model method helps to better manage inventory by understanding the behavioral patterns of users in the actual selection process (such as giving up certain products and choosing their competitors), and also allows supermarkets or suppliers to more accurately predict future sales trends.

[0034] In this embodiment, the sales forecasting model method further includes the following steps: Competing products and their prices, as well as products picked up by users and their prices, are sent to the suppliers of the products picked up by users. Suppliers can more flexibly judge the future sales of products and adjust production plans and inventory management more flexibly.

[0035] When a shopping cart is used by multiple users (such as parents and children), children may randomly take items from the shelf and put them into the shopping cart. However, these items may not meet the parents' purchasing intentions, and may even be put back on the shelf after being discovered by the parents within a short period of time. This situation leads to inaccurate results on the items taken out by the user, which in turn affects the accuracy of the sales prediction model method. See Figure 2 ,In this embodiment, the users corresponding to the cart include a first user and a second user, and the first user is a child; In step S102, when the commodities placed in the cart by the user are identified according to the photographed photos and a list of commodities placed in the cart by the user is obtained, the following steps are also included: Step S201, judging through the photographed photos whether the first user takes the goods from the display area of ​​the shelf and puts them into the trolley, if so, proceeding to step S202, judging whether the second user puts the goods back to the display area by himself or by letting the first user within a preset time, if so, proceeding to step S203, removing the goods put into the trolley by the first user from all the goods put into the trolley to obtain a list of goods put into the trolley by the users, if not, there are no goods put into the trolley by the first user, and all the goods are those that the second user wanted to buy.

[0036] Specifically, the first user (child) takes a toy from the shelf and puts it in the cart. The camera captures this action and preliminarily adds the toy to the product list. The second user (parent) notices the toy and decides not to let the child buy it within the first preset time (for example, 30 seconds, 40 seconds, or 50 seconds), so he asks the child to put the toy back on the shelf. By analyzing the photos taken by the camera, the behavior of the toy being put back is detected, and it is determined that it was done under the guidance of the second user. Finally, the toy is deleted and is not counted in the product list for subsequent analysis to avoid inaccurate product lists that affect the accuracy of competing products.

[0037] Although facial recognition technology can be used as a means to try to distinguish different users, due to the complexity and diversity of facial features, including but not limited to factors such as facial occlusion and age differences, the accuracy of user age recognition is not high. Figure 3 In this embodiment, the sales prediction model method identifies the first user and the second user through the following steps: Step S301, obtaining the pixel height marked in the captured photo; Step S302, obtaining the pixel height of the user in the taken photo; Step S303, obtaining the actual height of the marker, and obtaining the actual height of the user according to the pixel height of the marker, the pixel height of the user and the actual height of the marker, wherein the actual height of the user is recorded as h 1 , the user's pixel height is recorded as m 1 , the actual height of the mark is recorded as h 2 , the pixel height of the mark is recorded as m 2 , the user's actual height h 1 Calculated by the following formula:

[0038] Step S304, determine whether the actual height of the user is less than the height threshold (such as 1.0 meters, 1.1 meters or 1.2 meters). If so, proceed to step S305 to determine that the user is the first user. If not, proceed to step S306 to determine that the user is the second user. The second user refers to an adult or teenager with independent decision-making ability who serves as a decision maker in the shopping process. Compared with facial recognition, the height-based recognition method is less affected by environmental factors and has a higher accuracy rate. It improves the accuracy of data processing and operates without infringing the user's facial privacy.

[0039] In a further embodiment, the identification is a commodity label on the shelf, an end of a shelf layer, or an identification plate fixed on the layer or the ground. Since these identifications are usually fixed and will not move easily, the calculation error caused by the change of the reference object is reduced.

[0040] See also Figure 4 In this embodiment, in step S102, when identifying the commodities placed in the cart by the user according to the photographed photos, the following steps are also included: Step 401, identifying the commodity taken by the user from the shelf placement area based on the taken before and after photos; Step 402, identifying the product label corresponding to the display area according to the photographed photo, and obtaining the weight of the product according to the product label; Step 403, obtaining the weight difference detected by the pressure sensor on the placement area; Step 404 determines whether the weight obtained according to the product label and the weight difference are less than the weight threshold. If so, proceed to step 405 to determine whether the product taken out is correctly identified and output the product to the controller for use in step S104. If not, proceed to step 406 to use OCR technology or vision and deep learning technology to identify the correct name of the product taken out by the user from the shelf placement area and output it for use in step S104.

[0041] Specifically, OCR (Optical Character Recognition) technology is used to extract text information from images and convert it into a form that can be read by the processor. In a supermarket environment, OCR can be used to recognize text on product packaging, barcodes or information on labels, and perform post-processing steps such as grammar checking and spelling correction on the recognized text.

[0042] Specifically, vision and deep learning technologies can use deep learning frameworks (such as TensorFlow, PyTorch) to train convolutional neural networks (CNN) or other models suitable for image classification tasks, automatically extract complex features from images and classify or identify them. Supermarkets can use product templates in the database to train models in advance.

[0043] The camera is used to collect multiple video frames before and after the user takes out the product, and the target detection algorithm is used to identify whether the user has taken a product from the shelf. According to the shelf area where the removed product is located, the camera is controlled to automatically focus and shoot the product label (such as price, product description, etc.) corresponding to the placement area. These photos of the product labels that are not blocked by the user can be used. The standard weight of the product (for example: 3kg) is obtained. At the same time, the pressure sensor installed on the placement area records the weight difference before and after the product is taken (assuming it is 2.98kg, which is less than the weight threshold) to determine that the product is correctly identified. If the weight deviation is large (for example, the label shows 3kg, but the weighing difference is 1.5kg), the system starts OCR technology or combines vision and deep learning technology to re-identify the product to ensure that its name and specifications are accurate. Through the multi-dimensional product recognition mechanism, the high computing power requirements brought by direct reliance on OCR or deep learning technology can be reduced.

[0044] In some embodiments, when identifying the commodities placed in the cart by the user according to the photographed photos, the following steps are also included: The products in the photographs can be directly identified through OCR technology or vision and deep learning technology without combining with the above-mentioned weight judgment mechanism.

[0045] See also Figure 5 In this embodiment, the sales forecasting model method further includes the following steps: Step 501, counting the number of times the same product is taken out by users within a preset period and the corresponding competing products; Step 502, determine whether the frequency of occurrence of the competing product corresponding to the product taken out by the user is greater than the frequency threshold. If so, proceed to step 503, send a price comparison curve of the taken out product and its high-frequency competing products to the supplier of the taken out product. If not, store the information of the product taken out by the user and the competing products in the database.

[0046] Suppose that milk of a certain brand A is often put into the cart in the supermarket, but it is often replaced by milk of brand B or brand C at the final settlement. The processor counts the number of times the product has been taken out of the cart in the past month, and records whether other brands of milk were purchased as a substitute after each time it was taken out. The collected substitute products are counted for frequency, and it is found that brand B milk appears most frequently, exceeding the set threshold (such as more than 1,000 times per month). The processor automatically generates a price comparison curve between the product and brand B milk, showing the price difference between the two in different periods. The processor can also generate a frequency ranking chart of all competitive competing products. The above analysis results can be automatically sent to the supplier of brand A to remind them to pay attention to the market competition situation and make predictions for the next sales.

[0047] In this embodiment, the sales forecasting model method further includes the following steps: Get the past purchase date and sales amount of the items that the user has taken out from the cart; According to the sales data of the commodity, a plurality of preset date ranges and the preset activity intensity corresponding to each preset date range, a time series training data set is obtained, wherein the time series training data set includes features of the commodity and sales amounts corresponding to the features, wherein the features include identification features, date features, activity features and time series features; The model is trained using the loss function according to the time series training data set to obtain a sales amount prediction model; specifically, the loss function L is constructed based on the mean and standard deviation of the predicted sales amount, which is:

[0048] in, is the sales amount at time t The distribution of is the mean of the predicted sales amount, is the standard deviation of the predicted sales amount, i is an integer greater than or equal to 1 and less than or equal to N, and t is an integer greater than or equal to t 0 , and is an integer less than or equal to T.

[0049] The sales amount prediction model is used to predict the sales amount range of the goods taken out by the user from the cart. The sales amount range can be predicted based on the activity feature information of the goods, with low lag, which improves the accuracy of the prediction.

[0050] This embodiment also provides a sales forecasting model system based on multiple dimensions, which stores computer program instructions. When the computer program instructions are executed by a processor, the sales forecasting model method based on multiple dimensions as described in any of the above embodiments is implemented.

[0051] In this embodiment, the sales forecast model system includes a processor, a camera, a pressure sensor, and a supermarket database. The processor, as the control core of the entire system, is responsible for executing preset computer program instructions and coordinating data flows and task scheduling between modules. For example: receiving raw image data taken by the camera; accessing the supermarket database to obtain basic information of the product (such as product label, price, category, weight); controlling the sales module to output sales forecast results, competitive product analysis reports, etc., and sending feedback information to suppliers through an external communication interface, etc. The camera includes a high-definition wide-angle camera and an intelligent camera with edge computing capabilities, which transmits the video stream to the processor via a wired / wireless network. The camera can be deployed in important places in the supermarket according to actual needs to obtain a better field of view, such as above the shelf, near the channel and the settlement area. The video stream is transmitted to the processor via a wired / wireless network. The pressure sensor is used to monitor the weight change of the goods on the shelf, thereby assisting in identifying the goods that the user has taken or put back from the shelf. High-precision pressure sensors or weighing sensors (such as strain gauge sensors) are usually used. These sensors can accurately measure the weight of objects placed on them, convert the detected weight changes into electrical signals, and usually output them to the processor in the form of digital signals. Pressure sensors can be installed on different shelves or under specific commodity grids to independently monitor weight changes of commodities above them. The supermarket database stores basic commodity information, such as name, specification, price, category, weight, brand, etc. It can also provide commodity label images, barcodes, OCR feature libraries, etc. for commodity identification. The database is connected to the sales module and stores historical transaction orders.

[0052] Finally, it should be noted that although the above embodiments have been described in the specification and drawings of this application, this does not limit the scope of patent protection of this application. All technical solutions generated by replacing or modifying equivalent structures or equivalent processes based on the essential concept of this application using the contents recorded in the specification and drawings of this application, as well as directly or indirectly implementing the technical solutions of the above embodiments in other related technical fields, are included in the scope of patent protection of this application.

Claims

1. A sales forecasting model method based on multiple dimensions, characterized in that: The steps include: Take a photo of a user shopping in a supermarket using a shopping cart; Identify the items that the user has put in the cart based on the photographed photos, and obtain a list of the items that the user has put in the cart; Get the settlement order of the user's shopping; Compare the product information of the settlement order with the list of products that the user has put in the cart, and obtain the products that the user has taken out of the cart; Identify whether there are products of the same category in the settlement order based on the product taken out by the user. If so, obtain the products of the same category in the settlement order and use them as competing products of the product taken out by the user, obtain the selling price of the competing products from the product information of the settlement order, and obtain the corresponding selling price based on the product taken out by the user.

2. The sales forecasting model method according to claim 1, characterized in that: The users corresponding to the trolley include a first user and a second user, wherein the first user is a child; When the items placed in the cart by the user are identified according to the photographed photos and a list of items placed in the cart by the user is obtained, the following steps are also included: It is determined through the taken photos whether the first user takes the product from the display area of ​​the shelf and puts it into the trolley. If so, it is determined whether the second user puts the product back to the display area alone or with the help of the first user within a preset time. If so, the products placed in the trolley by the first user are removed from all the products placed in the trolley to obtain a list of the products placed in the trolley by the users.

3. The sales forecasting model method according to claim 2, characterized in that: The first user and the second user are identified by the following steps: Get the pixel height of the marker in the captured photo; Get the pixel height of the user in the captured photo; Get the actual height of the logo, and get the actual height of the user based on the pixel height of the logo, the pixel height of the user and the actual height of the logo; It is determined whether the actual height of the user is less than the height threshold. If so, the user is determined to be the first user; if not, the user is determined to be the second user.

4. The sales forecasting model method according to claim 3, characterized in that: The identification is a commodity label on the shelf, an end of a shelf layer, or an identification plate fixed on the layer or the ground.

5. The sales forecasting model method according to claim 1, characterized in that: When identifying the commodities placed in the cart by the user according to the photographed photos, the method further includes the following steps: Identify the products that the user has taken from the shelf placement area based on a number of photos taken before and after; Identify the product labels corresponding to the display area based on the photos taken, and obtain the weight of the product based on the product labels; Obtain the weight difference detected by the pressure sensor on the placement area; Determine whether the weight obtained according to the product label and the weight difference are less than the weight threshold. If so, it is determined that the product taken out is correctly identified. If not, use OCR technology or vision and deep learning technology to identify the correct name of the product taken out by the user from the shelf placement area and output it.

6. The sales forecasting model method according to claim 1, characterized in that: When identifying the commodities placed in the cart by the user according to the photographed photos, the method further includes the following steps: Identify products in photos using OCR technology or vision and deep learning technology.

7. The sales forecasting model method according to claim 1, characterized in that: The following steps are also included: The competing products and their selling prices, as well as the products picked up by the user and their selling prices are sent to the suppliers corresponding to the products picked up by the user.

8. The sales forecasting model method according to claim 1, characterized in that: The following steps are also included: Count the number of times the same product is selected by users within a preset period and the corresponding competing products; It is determined whether the frequency of occurrence of the competing product corresponding to the product selected by the user is greater than the frequency threshold. If so, a price comparison curve between the selected product and its high-frequency competing products is sent to the supplier of the product selected by the user.

9. The sales forecasting model method according to any one of claims 1 to 8, characterized in that: The following steps are also included: Get the past purchase date and sales amount of the items that the user has taken out from the cart; According to the sales data of the commodity, a plurality of preset date ranges and the preset activity intensity corresponding to each preset date range, a time series training data set is obtained, wherein the time series training data set includes features of the commodity and sales amounts corresponding to the features, wherein the features include identification features, date features, activity features and time series features; Performing model training using a loss function based on the time series training data set to obtain a sales amount prediction model; The sales amount prediction model is used to predict the sales amount range of the items that the user takes out from the cart.

10. A sales forecasting model system based on multiple dimensions, storing computer program instructions, characterized in that: When the computer program instructions are executed by a processor, the multi-dimensional sales forecasting model method is implemented as described in any one of claims 1 to 9.

Images