A food material cleaning parameter configuration method and device and electronic equipment

Abstract

The application discloses a food material cleaning parameter configuration method and device and electronic equipment. The method comprises the following steps: obtaining image information of food materials to be cleaned and a set of hardware capability parameters of a cleaning device; identifying the image information based on a preset multi-modal model to obtain a multi-dimensional identification result of the food materials to be cleaned; obtaining a cleaning parameter knowledge base; constructing a composite search condition based on a name identifier and a size classification, and matching a corresponding initial cleaning parameter combination in the cleaning parameter knowledge base; and verifying the initial cleaning parameter combination based on the set of hardware capability parameters to generate a target cleaning instruction. In the application, the limitation of traditional single category identification is broken through, the best cleaning parameter is adaptively matched in combination with the multi-dimensional features of food materials, the cleaning efficiency is improved, water resources are saved, and the parameter is ensured not to exceed the limit through hardware capability verification, so that the operation safety of the equipment is ensured.

Description

Technical Field

[0001] This application relates to the field of intelligent cleaning technology, and in particular to a method, apparatus and electronic device for configuring food cleaning parameters. Background Technology

[0002] Food cleaning, as a core preliminary step in food processing and daily meal preparation, plays a crucial role in removing pesticide residues, dirt particles, pathogenic microorganisms, and exogenous contaminants from the surface of food, directly impacting food safety and human health. With the rapid development of smart home and kitchen appliance technologies, smart fruit and vegetable sinks and cleaning machines with ultrasonic, ozone, and spray functions are becoming increasingly popular.

[0003] Existing smart cleaning solutions typically support simple interaction via mobile applications or control panels, where users manually select preset food categories, and the device uses the fixed process flow and water quality and temperature parameters hard-coded in the firmware to complete the cleaning.

[0004] However, existing cleaning solutions only focus on determining cleaning parameters based on the type of food, and cannot be flexibly adjusted according to the actual quantity, physical size and spatial distribution of the food. This results in different quantities of food of the same type having to use the same amount of water and processing time, which easily leads to the contradiction of wasting water resources in small-batch cleaning and insufficient cleaning in large-batch cleaning. Summary of the Invention

[0005] To address existing technical problems, this invention provides a method, apparatus, and electronic device for configuring food cleaning parameters. By acquiring food images and equipment hardware capability parameters, a cloud-based multimodal model is used to identify multidimensional food results. Based on this, composite retrieval conditions are constructed to match initial cleaning parameters in a cleaning parameter knowledge base. After hardware capability verification, instructions are generated to drive the cleaning equipment to execute. This method overcomes the limitations of traditional single-category recognition, combines multidimensional food features to adaptively match the optimal cleaning parameters, improves cleaning efficiency, saves water resources, and ensures that parameters do not exceed limits through hardware capability verification, thus guaranteeing equipment operation safety.

[0006] In a first aspect, embodiments of this application provide a method for configuring food cleaning parameters, the method comprising: Acquire image information of the food to be cleaned and a set of hardware capability parameters of the cleaning equipment; The image information is identified based on a preset multimodal model to obtain a multidimensional identification result of the food to be cleaned; the multidimensional identification result includes name identification and size classification; Access the cleaning parameter knowledge base; Based on name identifiers and size classifications, composite search conditions are constructed to match the corresponding initial cleaning parameter combinations in the cleaning parameter knowledge base. The initial cleaning parameter combination is verified based on the set of hardware capability parameters to generate a target cleaning instruction; the target cleaning instruction is used to drive the cleaning equipment to perform the cleaning operation.

[0007] In one optional embodiment, image information is identified based on a preset multimodal model to obtain a multidimensional identification result of the food to be cleaned, including: Based on a preset multimodal model, the image information is identified to obtain the visual reference objects and the names, types, quantities, and area proportions of the ingredients to be cleaned in the image information. The size classification of the ingredients to be cleaned is determined based on visual references, the quantity of ingredients, and the area ratio occupied by the ingredients. If there is only one type of ingredient, obtain the name and size classification of the ingredient to be cleaned; or if there are multiple types of ingredients, determine the main ingredient among the multiple ingredients to be cleaned based on the size classification, and use the size classification of the main ingredient as the size classification of the ingredients to be cleaned, and use the name and identification of the main ingredient as the name and identification of the ingredients to be cleaned; the size classification of the main ingredient is greater than the size classification of the other ingredients among the multiple ingredients to be cleaned. The multidimensional identification results are determined based on the name and size classification of the food items to be cleaned.

[0008] In one optional embodiment, determining the size classification of the food to be cleaned based on visual references, the quantity of food, and the area ratio occupied by the food includes: Obtain the reference physical dimensions of the visual reference object and the area proportion occupied by the reference object; The size conversion relationship between physical size and pixel area is determined based on the reference physical size and the area ratio occupied by the reference object; The pixel area of ​​each ingredient is determined based on the proportion of the area occupied by the ingredients and the quantity of ingredients. The physical dimensions of the ingredients are determined based on the conversion relationship between the pixel area and size of the ingredients. The size classification of the food to be cleaned is determined based on preset size thresholds and the physical size of the food.

[0009] In one optional embodiment, a composite search condition is constructed based on name identifier and size classification, and the corresponding initial cleaning parameter combination is matched in the cleaning parameter knowledge base, including: Combine name identifiers and size hierarchies to form a composite search key; If a first cleaning parameter matching the composite search key exists in the cleaning parameter knowledge base, the first cleaning parameter is determined as the initial cleaning parameter combination; or if a first cleaning parameter matching the composite search key does not exist in the cleaning parameter knowledge base, the type of food to be cleaned is determined, and a downgraded search is performed based on the type of food, name identifier, and size classification to match the corresponding initial cleaning parameter combination in the cleaning parameter knowledge base.

[0010] In one optional embodiment, a downgraded search is performed based on food type, name identifier, and size classification, and the corresponding initial cleaning parameter combination is matched in the cleaning parameter knowledge base, including: If a second cleaning parameter matching the name identifier exists in the cleaning parameter knowledge base, the second cleaning parameter is determined as the initial cleaning parameter combination; or if the second cleaning parameter does not exist in the cleaning parameter knowledge base, a search is performed based on the type and size of the food. Search based on food type and size classification, including: If a third cleaning parameter that matches the type and size classification of the food exists in the cleaning parameter knowledge base, the third cleaning parameter is determined as the initial cleaning parameter combination; or if a third cleaning parameter does not exist in the cleaning parameter knowledge base, a search is performed based on the type of food. Search based on food type, including: If a fourth cleaning parameter matching the type of food exists in the cleaning parameter knowledge base, the fourth cleaning parameter is determined as the initial cleaning parameter combination; or if a fourth cleaning parameter does not exist in the cleaning parameter knowledge base, a preset safe default cleaning parameter combination is obtained as the initial cleaning parameter combination.

[0011] In one optional embodiment, the multidimensional identification result further includes confidence level; obtaining a cleaning parameter knowledge base, including: Determine the first confidence level from the multidimensional recognition results; If the first confidence level is greater than or equal to the first preset threshold, retrieve the cleaning parameter knowledge base; or if the first confidence level is less than the first preset threshold but greater than or equal to the second preset threshold, send a result confirmation message; or if the first confidence level is less than the second preset threshold, send an image retake prompt message.

[0012] In one optional embodiment, the initial cleaning parameter combination is verified based on a set of hardware capability parameters to generate a target cleaning instruction, including: Obtain the target physical control parameters from the initial cleaning parameter combination and the corresponding equipment limit parameters from the hardware capability parameter set; If the target physical control parameter is greater than the equipment limit parameter, then the target physical control parameter will be clamped to the equipment limit parameter; The target cleaning command is generated based on the equipment's limit parameters; the target cleaning command carries a degradation flag. Send a parameter downgrade prompt message.

[0013] In one optional embodiment, after obtaining the cleaning parameter knowledge base, the process includes: If the cleaning parameter knowledge base fails to be retrieved, the basic parameter table stored locally is retrieved and used as the cleaning parameter knowledge base.

[0014] Secondly, embodiments of this application provide a food cleaning parameter configuration device, the device comprising: The first acquisition module is used to acquire image information of the food to be cleaned and a set of hardware capability parameters of the cleaning equipment. The image recognition module is used to recognize image information based on a preset multimodal model to obtain multidimensional recognition results of the food to be cleaned; the multidimensional recognition results include name identification and size classification; The second acquisition module is used to acquire the cleaning parameter knowledge base; The retrieval and matching module is used to construct composite retrieval conditions based on name identifiers and size classifications, and match the corresponding initial cleaning parameter combinations in the cleaning parameter knowledge base. The instruction generation module is used to verify the initial cleaning parameter combination based on the hardware capability parameter set and generate the target cleaning instruction; the target cleaning instruction is used to drive the cleaning equipment to perform the cleaning operation.

[0015] Thirdly, embodiments of this application provide an electronic device, which includes a processor and a memory. The memory stores at least one instruction, at least one program, code set, or instruction set. The processor loads and executes the at least one instruction, at least one program, code set, or instruction set to implement the food cleaning parameter configuration method of the first aspect.

[0016] Fourthly, embodiments of this application provide a computer-readable storage medium storing at least one instruction or at least one program, wherein the at least one instruction or at least one program is loaded and executed by a processor to implement the food cleaning parameter configuration method of the first aspect.

[0017] Fifthly, embodiments of this application provide a computer program product or computer program that includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the food cleaning parameter configuration method of the first aspect.

[0018] The food cleaning parameter configuration method, apparatus, and electronic device provided in this application have the following technical effects: This application acquires image information of the food to be cleaned and a set of hardware capability parameters for the cleaning equipment; it then uses a pre-defined multimodal model to recognize the image information, obtaining multidimensional recognition results of the food to be cleaned, including name identification and size classification; it acquires a cleaning parameter knowledge base; it constructs composite search conditions based on name identification and size classification, and matches the corresponding initial cleaning parameter combination in the cleaning parameter knowledge base; it verifies the initial cleaning parameter combination based on the hardware capability parameter set, generating a target cleaning instruction; and the target cleaning instruction drives the cleaning equipment to perform the cleaning operation. In this application, by acquiring food images and equipment hardware capability parameters, a cloud-based multimodal model is used to recognize multidimensional food results, and composite search conditions are constructed to match initial cleaning parameters in the cleaning parameter knowledge base. After hardware capability verification, an instruction is generated to drive the cleaning equipment to execute the operation. This overcomes the limitations of traditional single-category recognition, combines multidimensional food features to adaptively match the optimal cleaning parameters, improves cleaning efficiency and saves water resources, and ensures that parameters do not exceed limits through hardware capability verification, guaranteeing equipment operation safety. Attached Figure Description

[0019] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of an application environment provided in an embodiment of this application; Figure 2 This is a flowchart illustrating a method for configuring food cleaning parameters provided in an embodiment of this application. Figure 1 ; Figure 3 This is a flowchart illustrating a method for configuring food cleaning parameters provided in an embodiment of this application. Figure 2 ; Figure 4 This is a flowchart illustrating a method for configuring food cleaning parameters provided in an embodiment of this application. Figure 3 ; Figure 5 This is a schematic diagram of the structure of a food cleaning parameter configuration device provided in an embodiment of this application; Figure 6 This is a hardware structure block diagram of a server for a method of configuring food cleaning parameters provided in an embodiment of this application. Detailed Implementation

[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0022] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0023] Please see Figure 1 , Figure 1 This is a schematic diagram of an application environment provided in an embodiment of this application. The food cleaning parameter configuration system 100 includes a device terminal 101, a cloud 102, and a cleaning device 103. The device terminal 101 establishes communication connections with the cloud 102 and the cleaning device 103 respectively. The cloud 102 is the core data processing and parameter scheduling terminal, and the cleaning device 103 is the food cleaning execution terminal. The three work together to realize the intelligent and adaptive configuration of food cleaning parameters and automated cleaning operations.

[0024] In one possible embodiment, the device terminal 101 is preferably a smart portable terminal such as a smartphone or tablet. The device terminal 101 has a built-in dedicated control application (APP) and integrates an image acquisition module and a control module.

[0025] The image acquisition module is used to acquire raw image information of the food to be cleaned in the sink in real time; the control module is the core processing unit of the device terminal 101, responsible for completing the entire process control operation such as device binding, hardware parameter acquisition, image data uploading, parameter verification, instruction generation and issuance.

[0026] In one possible embodiment, the cloud 102 is equipped with a pre-trained multimodal model and a dynamically updated cleaning parameter knowledge base, and possesses core capabilities such as data inference, parameter retrieval, version management, canary release, and anomaly fallback. The cloud 102 can receive food image information uploaded by the device terminal 101, complete multi-dimensional feature recognition of the food through the multimodal model, and provide real-time parameter retrieval services to the device terminal 101. It also supports the addition, revision, rollback, and batch update of cleaning parameters without requiring upgrades or iterations of the terminal device.

[0027] In one possible embodiment, the cleaning device 103 is an intelligent fruit and vegetable washing tank machine, which integrates an ultrasonic cleaning module, an ozone generation module, a water level adjustment module, a water temperature control module, and a spray module. It can receive target cleaning commands from the device terminal 101 and accurately execute corresponding cleaning actions such as ultrasonic power, ozone concentration, water level and volume, and cleaning duration. Simultaneously, the cleaning device 103 has hardware parameter feedback capabilities, allowing it to synchronize its own hardware performance limits with the device terminal 101.

[0028] In this embodiment, the food cleaning parameter configuration system 100 acquires image information of the food to be cleaned and a set of hardware capability parameters of the cleaning equipment through the device terminal 101, and uploads the image information to the cloud 102. The cloud 102 recognizes the image information based on a preset multimodal model, acquires and returns multidimensional recognition results of the food to be cleaned to the device terminal 101; the multidimensional recognition results include name identification and size classification. The device terminal 101 acquires the cleaning parameter knowledge base of the cloud 102; constructs composite search conditions based on name identification and size classification, matches the corresponding initial cleaning parameter combination in the cleaning parameter knowledge base; verifies the initial cleaning parameter combination based on the set of hardware capability parameters, and generates a target cleaning instruction; the target cleaning instruction is used to send to the cleaning equipment 103 to drive the cleaning equipment 103 to perform the cleaning operation.

[0029] In this application, by acquiring food images and equipment hardware capability parameters, multi-dimensional recognition results are obtained using a preset multimodal model in the cloud 102. Based on this, composite retrieval conditions are constructed to match initial cleaning parameters in the cleaning parameter knowledge base. After hardware capability verification, instructions are generated to drive the cleaning equipment 103 to execute. This breaks through the limitations of traditional single-category recognition, combines multi-dimensional food features to adaptively match the best cleaning parameters, improves cleaning efficiency and saves water resources, and ensures that parameters do not exceed limits through hardware capability verification, thus guaranteeing the safe operation of the equipment.

[0030] The following describes a specific embodiment of a method for configuring food cleaning parameters according to this application. Figure 2 This is a flowchart illustrating a method for configuring food cleaning parameters provided in an embodiment of this application. Figure 1This specification provides method operation steps as shown in the embodiments or flowcharts, but based on conventional or non-inventive labor, more or fewer operation steps may be included. The order of steps listed in the embodiments is merely one possible execution order among many and does not represent the only execution order. In actual system or server products, the methods shown in the embodiments or drawings can be executed sequentially or in parallel (e.g., in a parallel processor or multi-threaded processing environment). Specifically, as shown in the embodiments or drawings... Figure 2 As shown, it may include: S201: Acquire image information of the food to be cleaned and a set of hardware capability parameters of the cleaning equipment.

[0031] S202: Based on a preset multimodal model, image information is identified to obtain multidimensional identification results of the food to be cleaned; the multidimensional identification results include name identification and size classification.

[0032] S203: Obtain the cleaning parameter knowledge base.

[0033] S204: Construct composite search conditions based on name identifiers and size classifications, and match the corresponding initial cleaning parameter combinations in the cleaning parameter knowledge base.

[0034] S205: Verify the initial cleaning parameter combination based on the hardware capability parameter set, and generate a target cleaning instruction; the target cleaning instruction is used to drive the cleaning equipment to perform the cleaning operation.

[0035] Figure 3 This is a flowchart illustrating a method for configuring food cleaning parameters provided in an embodiment of this application. Figure 2 The method may include: S301: Acquire image information of the food to be cleaned and a set of hardware capability parameters of the cleaning equipment.

[0036] In this embodiment of the application, the user completes the Bluetooth or WiFi binding of the cleaning equipment through the dedicated application (APP) of the device terminal. After the device is successfully bound, the device terminal automatically reads and caches the hardware capability parameter set (Cap) of the cleaning equipment.

[0037] The hardware capability parameter set Cap is structured data, specifically including hardware limit parameters such as whether it supports ozone function, whether it supports ultrasonic cleaning function, whether it supports water temperature heating function, whether it supports spray function, maximum ultrasonic power, maximum ozone concentration, highest supported water level, and lowest working water level, which fully cover all performance boundaries of the cleaning equipment.

[0038] Meanwhile, users can evenly lay the food to be cleaned inside the water tank of the cleaning equipment or on a special shooting mat, and the image acquisition module of the equipment terminal will complete a single image capture to obtain clear image information of the food to be cleaned.

[0039] S302: Based on a preset multimodal model, image information is identified to obtain multidimensional recognition results of the food to be cleaned.

[0040] In this embodiment, the device terminal uploads the collected food image information to the cloud in real time. The cloud calls a preset multimodal large model to perform intelligent reasoning and recognition on the image information, and obtains and returns the multidimensional recognition results of the food to be cleaned to the device terminal.

[0041] Unlike traditional single-vision recognition schemes, the multimodal model in this application can extract multidimensional features of food ingredients and output standardized and structured multidimensional recognition results.

[0042] In one optional embodiment, the multidimensional identification results mainly include the name identification and size classification of the ingredients.

[0043] In an optional embodiment, step S302 involves recognizing image information based on a preset multimodal model to obtain a multidimensional recognition result of the food to be cleaned, including: S3021: Based on a preset multimodal model, image information is identified to obtain the visual reference objects and the names, types, quantities, and area proportions of the ingredients to be cleaned from the image information.

[0044] In this embodiment, after the device terminal uploads the collected food image information to the cloud, the cloud performs global feature reasoning and analysis on the entire image information through a preset multimodal model, and completes the synchronous recognition and output of multi-dimensional features of the calibrated object and food in one go.

[0045] The specific identification content includes: visual reference objects within the image, which can be calibrated objects with known standard physical dimensions, such as the edge of the sink wall, the outline of the special shooting pad, or the screen of the device terminal; at the same time, it accurately identifies the multi-dimensional basic features corresponding to the food to be cleaned, including the name of the food, the specific type of food, the number of each type of food, and the proportion of pixel area occupied by each type of food in the image field of view.

[0046] S3022: Determine the size classification of the food to be cleaned based on visual references, the quantity of food, and the area ratio occupied by the food.

[0047] The size grading is specifically divided into three levels: S (small), M (medium), and L (large).

[0048] S3023: Determine if there is only one type of ingredient. If yes, proceed to S3024; otherwise, proceed to S3025.

[0049] S3024: Obtain the name and size classification of the food to be cleaned.

[0050] In one optional embodiment, if there is only one type of food ingredient, the name identifier and the corresponding calculated size grade of the unique food ingredient to be cleaned are directly extracted and used as the basic characteristic parameters of the food ingredient for this cleaning task, without the need for multi-food screening and priority determination.

[0051] S3025: Identifying the main ingredient from a variety of ingredients to be cleaned based on size grading.

[0052] S3026: The size classification of the main ingredient is determined as the size classification of the ingredient to be cleaned, and the name identifier of the main ingredient is determined as the name identifier of the ingredient to be cleaned.

[0053] In another optional embodiment, if there are multiple types of ingredients, the main ingredient is determined from the multiple ingredients to be cleaned based on size grading, the size grading of the main ingredient is determined as the size grading of the ingredients to be cleaned, and the name identifier of the main ingredient is determined as the name identifier of the ingredients to be cleaned.

[0054] In this embodiment, the size grade of the main ingredient is greater than that of the other ingredients among the various ingredients to be cleaned. Specifically, by comparing the size grades of all ingredients, the ingredient with the highest size grade and the largest overall volume is selected as the main ingredient for this mixed cleaning task. The size grade of this main ingredient is greater than that of all other auxiliary ingredients, indicating that it has the highest cleaning load, water volume requirement, and cleaning time requirement.

[0055] This application adopts a main ingredient convergence strategy that prioritizes the largest volume, ensuring that the final matched cleaning parameters can meet the cleaning requirements of the ingredients with the highest load, and avoiding the problem of incomplete cleaning and substandard cleaning effect of large-volume ingredients during the cleaning process of mixed ingredients.

[0056] S3027: Determine multidimensional identification results based on the name and size classification of the food to be cleaned.

[0057] In this embodiment of the application, after completing feature convergence for single or multiple food ingredients, the unique name and size classification of the food ingredient to be cleaned are finally determined. Combined with the synchronously output recognition confidence and food category information, a structured and standardized multidimensional food ingredient recognition result is generated, which facilitates subsequent retrieval.

[0058] In an optional embodiment, step S3022 determines the size classification of the food to be cleaned based on visual references, the quantity of food, and the area ratio occupied by the food, including: S30221: Obtain the reference physical dimensions of the visual reference object and the area ratio occupied by the reference object.

[0059] S30222: Determine the size conversion relationship between physical size and pixel area based on the reference physical size and the area ratio occupied by the reference object.

[0060] S30223: Determine the pixel area of ​​the food ingredient based on the area ratio and quantity of the food ingredient.

[0061] S30224: Determine the physical size of the food ingredient based on the conversion relationship between the pixel area and size of the food ingredient.

[0062] S30225: Determine the size classification of the food to be cleaned based on the preset size threshold and the physical size of the food.

[0063] In this embodiment, the size classification is determined based on the food type, quantity, area ratio of the food, and visual reference objects in the image information identified by a preset multimodal model. By combining the overall area ratio of the food in the image, the number of individual food items, and the stacked volume of the food, and using fixed visual reference objects such as the edge of the sink and the size of the device terminal screen as calibration benchmarks, a precise size classification result is obtained through comprehensive calculation, which can accurately reflect the overall volume and weight of the food to be cleaned.

[0064] In one possible embodiment, the step of determining the size classification specifically includes: First, a pre-defined multimodal model identifies fixed visual reference objects in the image information. These reference objects can be selected from fixed calibration objects such as the inner wall contour of a water tank, the outer contour of a dedicated shooting pad, or a terminal screen of known standard size. The model identifies the area ratio occupied by the reference object, which is the pixel length, width, and area of ​​the visual reference object in the image. The model obtains the factory-preset reference physical size of the visual reference object, establishes a one-to-one correspondence between pixel coordinates and real physical space, and generates a globally unified size conversion relationship. This achieves accurate conversion between physical size and pixel area, eliminating scale errors caused by shooting distance and shooting angle.

[0065] Secondly, the pixel area of ​​each ingredient is determined based on its area ratio and quantity. The physical size of the ingredients is then determined based on the pixel area and size conversion relationship. The model calculates the total pixel area of ​​the entire captured field of view, and combines this with the overall area ratio of the identified ingredients. The pixel area of ​​this type of ingredient is calculated using the formula: "Total pixel area of ​​ingredients = Total pixel area of ​​the image × Area ratio of ingredients." The total pixel area of ​​this type of ingredient is then determined based on its quantity. Furthermore, relying on the size conversion relationship established using the aforementioned visual references, the pixel area of ​​each ingredient is converted into its physical size.

[0066] In one possible embodiment, the dimensions can be divided according to the area, or the volume can be calculated from the area and then the dimensions can be divided according to the volume.

[0067] Finally, the system has a preset standardized size threshold. The calculated physical size of the food is matched with the preset size threshold. Based on the matching result, the food to be cleaned is divided into corresponding S, M, and L size grades, providing accurate volume basis for subsequent adaptive cleaning parameter retrieval.

[0068] In this embodiment, the following example illustrates how to determine the name and size classification if there are multiple ingredients: A large portion of spinach and a small amount of strawberries are mixed and washed together. During image acquisition, a rectangular metal frame at the edge of the water tank was captured as a fixed visual reference. The standard physical dimensions of the water tank are 400mm × 350mm. Multimodal model detection determined that the pixel dimensions of the rectangular frame in the image are 1600px × 1400px. By converting the pixel dimensions to the physical dimensions, a fixed scale mapping relationship was generated: 4px corresponds to 1mm, and the area per unit pixel corresponds to 0.0625mm² of the actual physical area, thus completing the calibration of the size conversion relationship.

[0069] The model accurately extracted all food ingredient regions through semantic segmentation, calculating the total pixel area of ​​the ingredients to be 480,000 px². After scaling, the actual physical area occupied by the ingredients in the sink was determined to be 30,000 mm², representing 21.4% of the effective washing area at the bottom of the sink. The model also identified two types of ingredients: loosely packed, numerous spinach leaves (a leafy green), and four scattered strawberries (a berry). Combining the density models for loose leafy greens and berries, the total volume of the spinach was estimated to be approximately 1800 cm³, and the total volume of the strawberries approximately 120 cm³.

[0070] The preset size thresholds for this application are: S grade (small portion) with a volume of less than 500cm³, M grade (medium portion) with a volume between 500cm³ and 1200cm³, and L grade (large portion) with a volume greater than 1200cm³. Therefore, the size classification of spinach is L (large portion), and the size classification of strawberries is S (small portion).

[0071] The model compares the size grading of the two types of ingredients. The volume of spinach in grade L is significantly larger than that of strawberries in grade S. Therefore, spinach is selected as the main ingredient for this cleaning. L (large portion) is used as the size grading and "spinach" is used as the name identifier. Then, "spinach|L" can be used as a composite search condition for subsequent cleaning parameter matching.

[0072] In one optional embodiment, the multidimensional recognition result also includes a confidence score, which is used to characterize the accuracy of the food identification result. The value range is 0 to 1, and the higher the value, the stronger the reliability of the recognition result.

[0073] S303: Obtain the cleaning parameter knowledge base.

[0074] In this embodiment, after the device terminal completes the food identification, it needs to combine the identification confidence level and the terminal network status to dynamically select the source of the cleaning parameter knowledge base, and at the same time complete the validity verification of the identification result, which is specifically divided into a confidence level verification step and a network adaptation verification step.

[0075] In one optional embodiment, the confidence verification step, namely obtaining the cleaning parameter knowledge base, includes: S3031: Determine the first confidence level in the multidimensional recognition results.

[0076] In this embodiment of the application, if there are multiple ingredients in the image information, the first confidence level is the highest confidence level among them.

[0077] S3032: Determine whether the first confidence level is greater than or equal to the first preset threshold. If yes, execute S3033; otherwise, execute S3034.

[0078] S3033: Obtain the cleaning parameter knowledge base.

[0079] S3034: Determine whether the first confidence level is greater than or equal to the second preset threshold. If yes, execute S3035; otherwise, execute S3036.

[0080] S3035: Send result confirmation message.

[0081] S3036: Send image retake prompt message.

[0082] In one possible embodiment, if the first confidence level is greater than or equal to the first preset threshold, the identification result is determined to be accurate and effective, and the latest cleaning parameter knowledge base is directly obtained.

[0083] In one possible embodiment, if the first confidence level is less than the first preset threshold and greater than or equal to the second preset threshold, it is determined that there is a slight uncertainty in the current identification result, and a result confirmation message is sent. Specifically, the device terminal pops up an interactive pop-up window, sends a result confirmation message to the user, prompts the user to check the food identification result, and after the user confirms that it is correct, the parameter retrieval process continues.

[0084] In another possible embodiment, if the first confidence level is less than the second preset threshold, it is determined that the recognition result is seriously distorted and has extremely low reliability. The device terminal blocks all subsequent cleaning processes, sends an image retake prompt to the user, guides the user to rearrange the ingredients and take images, until the recognition result meets the accuracy requirements.

[0085] In one optional embodiment, the network adaptation verification step, namely obtaining the cleaning parameter knowledge base, includes: S3031: Determine whether the terminal device has established a communication connection with the cloud. If yes, execute S3032; otherwise, execute S3033.

[0086] S3032: Obtain the cleaning parameter knowledge base.

[0087] S3033: Obtain the basic parameter table stored locally on the terminal device.

[0088] S3034: Define the basic parameter table as the cleaning parameter knowledge base.

[0089] In one alternative embodiment, a cleaning parameter knowledge base is obtained after the terminal device establishes a communication connection with the cloud.

[0090] In another optional embodiment, if the terminal device cannot establish a communication connection with the cloud and the cleaning parameter knowledge base fails to be obtained, specifically in an abnormal state such as network outage or excessive network latency, the cloud knowledge base call is abandoned, and the basic parameter table stored locally on the terminal device is obtained and the basic parameter table is determined as the cleaning parameter knowledge base.

[0091] In this embodiment, the basic parameter table pre-stores standard cleaning parameters for two sizes (large and small) of 20 frequently used food ingredients, ensuring normal device operation even during network outages. Simultaneously, the device continuously monitors the network status and proactively prompts the user to switch to the latest and optimal parameters from the cloud once the network is restored.

[0092] In other possible embodiments, a lightweight edge-side multimodal small model can also be deployed locally on the device terminal to complete food image recognition offline when the device terminal cannot establish a communication connection with the cloud, and output two core multidimensional recognition results: food name identification and size classification, to make up for the lack of availability of large cloud models in the offline scenario.

[0093] In this embodiment, the food images captured by the user are not uploaded to the cloud, but are directly sent to the on-device small model locally. The on-device small model has built-in lightweight semantic segmentation and object detection branches, which independently identify visual reference objects such as the edge of the sink and the shooting pad in the image, and simultaneously extract visual features such as the food area, the number of individual food items, and the proportion of food area in the image. No cloud-based image analysis is required, and the specific recognition process is similar to that of the preset multimodal large model.

[0094] The device terminal continuously polls and monitors the cloud communication status. Once the network returns to normal, it automatically prompts the user that "this time we are using the local offline small model for recognition. After connecting to the network, you can take a new picture to obtain the cloud large model for high-precision recognition and optimal cleaning parameters." The user can choose to retake the image and switch to the cloud large model recognition link to obtain a cloud parameter solution with higher accuracy and wider coverage of ingredients.

[0095] This embodiment achieves fully offline food category recognition and food size grading by deploying a lightweight model on the terminal. Even when offline, it can still adaptively match differentiated cleaning parameters based on real-shot images, instead of using a one-size-fits-all approach. Furthermore, the lightweight model is compressed and optimized, consuming minimal terminal storage and computing resources, preventing app lag and excessive power consumption, thus balancing offline usability with smooth terminal operation.

[0096] S304: Construct composite search conditions based on name identifiers and size classifications, and match the corresponding initial cleaning parameter combinations in the cleaning parameter knowledge base.

[0097] In this embodiment, to achieve precise matching of both ingredient category and quantity, the traditional single-category search method is abandoned. Instead, a composite search logic of "name identifier | size classification" is adopted to accurately match differentiated initial cleaning parameter combinations. Specific details regarding the classification search and fallback process will be provided later. Figure 4 Corresponding implementation examples.

[0098] S305: Verify the initial cleaning parameter combination based on the hardware capability parameter set, and generate the target cleaning instruction.

[0099] In this embodiment, after the device terminal obtains the initial cleaning parameter combination matched by the cloud, it does not directly issue the execution. Instead, it combines the hardware capability parameter set of the cleaning device to complete the compatibility verification, solve the parameter adaptation problem of different configuration devices, avoid the device failure caused by parameter exceeding the limit, and finally generate a target cleaning instruction that can be safely executed, and issue it to the cleaning device to drive it to complete the cleaning operation.

[0100] Specifically, the initial cleaning parameter combination is verified based on the hardware capability parameter set to generate the target cleaning instruction, including: S3051: Obtain the target physical control parameters and the corresponding equipment limit parameters from the set of hardware capability parameters in the initial cleaning parameter combination.

[0101] In one optional embodiment, the target physical control parameters in the initial cleaning parameter combination are extracted, specifically including ultrasonic power, ozone concentration, cleaning water level, cleaning time, spray intensity, etc., and the corresponding equipment limit parameters in the hardware capability parameter set are matched and retrieved, that is, the maximum support threshold of each physical parameter.

[0102] S3052: Determine whether the target physical control parameter is greater than the equipment limit parameter. If yes, execute S3053; otherwise, execute S3056.

[0103] In the embodiments of this application, each target physical control parameter is compared with the corresponding equipment limit parameter to determine whether the target physical control parameter is greater than the equipment limit parameter.

[0104] S3053: Clamp the target physical control parameters to the equipment limit parameters.

[0105] S3054: Generate target cleaning instructions based on equipment limit parameters.

[0106] S3055: Send parameter downgrade prompt message.

[0107] S3056: Generate target cleaning instructions based on target physical control parameters.

[0108] In one optional embodiment, if the target physical control parameter is greater than the device's limit parameter, it is determined that the parameter exceeds the device's hardware carrying capacity. The target physical control parameter is then adaptively clamped to the device's corresponding limit parameter to prevent overload operation. After the parameter clamping is completed, a target cleaning instruction is generated based on the corrected device limit parameter. At the same time, a degradation mark is carried in the instruction data packet, recording the original parameter value, the value after clamping, and the degradation field information. The device terminal synchronously sends parameter degradation prompt information to the user, clearly informing the user of the specific degradation parameter type, the original recommended parameter, and the device adaptation parameter, so that the user can know the device's operating status in real time.

[0109] In another optional embodiment, if the target physical control parameters are greater than the device limit parameters, it is determined that the parameters are fully compatible with the device hardware capabilities and do not need to be corrected. Standard target cleaning instructions are directly generated based on the original target physical control parameters.

[0110] Figure 4 This is a flowchart illustrating a method for configuring food cleaning parameters provided in an embodiment of this application. Figure 3 In one optional embodiment, a composite search condition is constructed based on name identifier and size classification, and the corresponding initial cleaning parameter combination is matched in the cleaning parameter knowledge base, including: S401: Combine name identifier and size hierarchy to form a composite search key.

[0111] In this embodiment, the device terminal concatenates the food name identifier and size classification output by the multimodal model to construct a unique composite search key. The search key format is uniformly "food name|size classification", for example, "spinach|L" represents a large portion of spinach and "strawberry|S" represents a small portion of strawberry. The structured composite key achieves dual positioning of category and quantity, providing support for accurate retrieval.

[0112] S402: Determine whether the first cleaning parameter exists in the cleaning parameter knowledge base. If yes, execute S403; otherwise, execute S404.

[0113] S403: Determine the first cleaning parameter as the initial cleaning parameter combination.

[0114] S404: Determine the type of food to be cleaned.

[0115] In this embodiment of the application, the major categories of the food to be cleaned are determined based on the multidimensional recognition results, such as leafy greens, mushrooms, root vegetables, and seafood.

[0116] S405: Perform a downgraded search based on food type, name, and size classification, and match the corresponding initial cleaning parameter combination in the cleaning parameter knowledge base.

[0117] In one optional embodiment, if a first cleaning parameter that matches the composite search key exists in the cleaning parameter knowledge base, the first cleaning parameter is directly determined as the initial cleaning parameter combination. This parameter is the optimal cleaning parameter that is suitable for the current food category and volume, and has the best cleaning accuracy and water-saving effect.

[0118] In another optional embodiment, if there is no first cleaning parameter matching the composite search key in the cleaning parameter knowledge base, it is determined that the current ingredient is a newly added ingredient or an uncommon ingredient. A multi-level downgrade search mechanism is immediately initiated to gradually expand the search scope and ensure effective parameter matching. Specifically, the ingredient type of the ingredient to be cleaned is determined, and a downgrade search is performed based on the ingredient type, name identifier, and size classification. The corresponding initial cleaning parameter combination is matched in the cleaning parameter knowledge base.

[0119] In one optional embodiment, a downgraded search is performed based on food type, name identifier, and size classification, and the corresponding initial cleaning parameter combination is matched in the cleaning parameter knowledge base, including: S4051: Determine whether a second cleaning parameter exists in the cleaning parameter knowledge base. If yes, execute S4052; otherwise, execute S4053.

[0120] Specifically, it determines whether there is a second cleaning parameter in the cleaning parameter knowledge base that matches the ingredient name identifier. Size classification is not limited. The second cleaning parameter is the cleaning parameter whose size is closest to the ingredient name identifier.

[0121] S4052: Determine the second cleaning parameter as the initial cleaning parameter combination.

[0122] S4053: Search based on food type and size classification.

[0123] In one optional embodiment, if a second cleaning parameter matching the name identifier exists in the cleaning parameter knowledge base, the second cleaning parameter is determined as the initial cleaning parameter combination, and the parameter whose size grade is closest to the current food volume is selected as the optimal parameter and determined as the initial cleaning parameter combination.

[0124] In another optional embodiment, if the second cleaning parameter is not found in the cleaning parameter knowledge base, the search logic jumps to the next level and searches are performed based on the type and size of the food.

[0125] In one optional embodiment, the retrieval is based on food type and size classification, including: S4054: Determine whether a third cleaning parameter exists in the cleaning parameter knowledge base. If yes, execute S4055; otherwise, execute S4056.

[0126] Specifically, it determines whether a third cleaning parameter exists in the cleaning parameter knowledge base that matches the food category and current size classification.

[0127] S4055: Determine the third cleaning parameter as the initial cleaning parameter combination.

[0128] S4056: Search based on food type.

[0129] In one optional embodiment, if a third cleaning parameter exists in the cleaning parameter knowledge base that matches the type and size classification of the food, the third cleaning parameter is determined as the initial cleaning parameter combination to ensure the cleaning adaptability of the same type of food.

[0130] In another optional embodiment, if the third cleaning parameter is not found in the cleaning parameter knowledge base, the search logic jumps to the next level and searches based solely on the type of food ingredient.

[0131] In one optional embodiment, the search is based on the type of food ingredient, including: S4057: Determine whether the fourth cleaning parameter exists in the cleaning parameter knowledge base. If yes, execute S4058; otherwise, execute S4059.

[0132] Specifically, it determines whether a fourth cleaning parameter matching the food category exists in the cleaning parameter knowledge base, without limiting the size classification.

[0133] S4058: Determine the fourth cleaning parameter as the initial cleaning parameter combination.

[0134] S4059: Obtain the preset safe default cleaning parameter combination as the initial cleaning parameter combination.

[0135] In one optional embodiment, if a fourth cleaning parameter matching the type of food exists in the cleaning parameter knowledge base, the fourth cleaning parameter is determined as the initial cleaning parameter combination.

[0136] In another optional embodiment, if the fourth cleaning parameter is not found in the cleaning parameter knowledge base, and if no matching parameter is found at any of the above search levels, a preset safe default cleaning parameter combination is obtained as the initial cleaning parameter combination. Specifically, the default parameters are: water level 60mm, ultrasonic power 20W, ozone concentration 0%, and cleaning time 120s. These standardized safety parameters serve as a safety net to ensure stable equipment operation and basic food cleaning effectiveness.

[0137] To address the shortcomings of existing technologies, such as cleaning parameters being fixed in firmware and application code, iterative updates relying on over-the-air (OTA) upgrades and version updates, high costs of synchronizing parameters for existing equipment, and lagging adaptation to food safety standards, this application adopts a cloud-based architecture that decouples knowledge from terminal code. All food cleaning process parameters and food adaptation rules are uniformly stored in a cloud-based, maintainable knowledge base, freeing them from the constraints of device firmware and local application (APP) code. This enables online expansion, real-time revision, and painless iterative updates of cleaning parameters without requiring any version upgrades or firmware updates for existing cleaning equipment or user terminal APPs.

[0138] In this embodiment, the cloud-based cleaning parameter knowledge base adopts an independently deployed cloud architecture, completely decoupled from the terminal control logic and device execution logic. It employs a key-value pair structured storage model. The index key of the knowledge base is uniformly designed as a composite string of "ingredient name | size grade," and the stored value is a standardized and complete combination of cleaning parameters for the corresponding ingredient and size. Each parameter record fully covers all controllable hardware parameters such as ultrasonic power, ozone concentration, cleaning water level, water inflow, cleaning duration, and spray level. This storage structure perfectly matches the aforementioned multimodal recognition results and composite retrieval logic, enabling accurate retrieval and rapid access.

[0139] In one possible implementation, the cloud-based knowledge base is equipped with complete backend operation and maintenance management capabilities, supporting vendors to continuously iterate and optimize the cleaning parameter system, specifically including several core operation and maintenance functions: The first feature is the addition of new entries. Manufacturers can add exclusive cleaning parameter entries for niche ingredients and agricultural product categories through the management backend as needed, while adapting to differentiated parameter configurations for different size grades to continuously expand the range of ingredients covered.

[0140] Secondly, there is a parameter revision function. Manufacturers can iteratively revise the cleaning parameters corresponding to existing food ingredients based on the latest food safety cleaning standards, a large amount of experimental test data, and user feedback data, thereby optimizing the cleaning process.

[0141] Third is the gray-scale release function. For new and revised parameters, a small-scale user gray-scale verification mechanism is adopted. The parameters are first pushed to a portion of terminal devices to complete the actual test. After verifying that the cleaning effect is stable and the device operation is normal, the full release is then carried out to avoid the risks of batch updates.

[0142] Fourth is the version management and rollback function. Every time a parameter is added or revised, a unique version number, modification time, effective time and modified content are automatically recorded. It supports one-click rollback of parameter versions in abnormal scenarios to ensure the stability of system operation.

[0143] In one possible embodiment, this application also employs a real-time cloud-based retrieval mechanism to ensure parameter timeliness, completely abandoning the model of caching and fixing parameters locally on the terminal. Each time a user initiates the food cleaning process and completes food image recognition, the device terminal proactively initiates a real-time query request to the cloud knowledge base to dynamically obtain the latest version of the food cleaning parameter combination. The device terminal does not permanently store or cache cloud knowledge base entries locally; it only temporarily caches the results of a single query for the generation of the cleaning instruction, ensuring that the parameters used for each food cleaning are the latest iterations from the manufacturer and the optimal parameters that comply with the latest food safety standards. This fundamentally solves the problem of outdated parameters and obsolete processes in existing equipment.

[0144] From a technical perspective, this solution completely decouples the food cleaning parameter knowledge layer from the terminal code layer and device firmware layer, changing the traditional hard-coded and fixed technical architecture. It centralizes parameter iteration capabilities entirely in the cloud, requiring only cloud backend data maintenance for parameter updates, without modifying the terminal program or device firmware, significantly reducing parameter iteration costs and existing equipment maintenance costs. This solution supports unlimited expansion of food categories, continuously adapting to the cleaning needs of new and niche ingredients, and quickly keeping pace with food safety standard iterations and process optimizations, updating optimal cleaning parameters in real time. This application eliminates user-end upgrade operations, seamlessly completing device parameter updates. Through a dual-mode design of real-time cloud retrieval and local offline backup, it balances parameter real-time performance and system operational stability.

[0145] This application achieves adaptive configuration of food cleaning parameters through the aforementioned technical solution. Compared to traditional technologies, it relies on multimodal visual recognition to replace manual selection, offering convenient operation, higher recognition accuracy, and coverage of both niche and rare food ingredients. By combining the matching parameters with both the food category and volume, it solves the problems of incomplete cleaning and water waste caused by fixed parameters. Furthermore, it enables decoupled cloud-based iteration of cleaning parameters, allowing parameter updates without requiring device firmware or application upgrades. Finally, through hardware parameter verification and adaptive clamping mechanisms, it adapts to cleaning equipment with different configurations, completely avoiding the risk of equipment exceeding its limits and balancing cleaning effectiveness, equipment safety, and ease of iteration.

[0146] This application also provides a device for configuring food washing parameters. Figure 5 This is a schematic diagram of the structure of a food cleaning parameter configuration device provided in an embodiment of this application, as shown below. Figure 5 As shown, the device 500 includes: The first acquisition module 501 is used to acquire image information of the food to be cleaned and a set of hardware capability parameters of the cleaning equipment. The image recognition module 502 is used to recognize image information based on a preset multimodal model to obtain multidimensional recognition results of the food to be cleaned; the multidimensional recognition results include name identification and size classification; The second acquisition module 503 is used to acquire the cleaning parameter knowledge base; The retrieval matching module 504 is used to construct composite retrieval conditions based on name identifier and size classification, and match the corresponding initial cleaning parameter combination in the cleaning parameter knowledge base; The instruction generation module 505 is used to verify the initial cleaning parameter combination based on the hardware capability parameter set and generate a target cleaning instruction; the target cleaning instruction is used to drive the cleaning equipment to perform the cleaning operation.

[0147] In an optional embodiment, it further includes: The first recognition module is used to recognize image information based on a preset multimodal model, and obtain the name, type, quantity, and area ratio of the visual reference objects and the food to be cleaned in the image information. The first determining module is used to determine the size classification of the food to be cleaned based on visual reference objects, the quantity of food, and the area ratio occupied by the food. The second determining module is used to obtain the name identifier and size classification of the ingredient to be cleaned if there is only one type of ingredient; or if there are multiple types of ingredients, determine the main ingredient among the multiple ingredients to be cleaned based on the size classification, determine the size classification of the main ingredient as the size classification of the ingredient to be cleaned, and determine the name identifier of the main ingredient as the name identifier of the ingredient to be cleaned; the size classification of the main ingredient is greater than the size classification of other ingredients among the multiple ingredients to be cleaned besides the main ingredient. The third determination module is used to determine the multidimensional identification results based on the name and size classification of the food to be cleaned.

[0148] In an optional embodiment, it further includes: The third acquisition module is used to acquire the reference physical size of the visual reference object and the area ratio occupied by the reference object; The third determining module is used to determine the size conversion relationship between physical size and pixel area based on the reference physical size and the area ratio occupied by the reference object; The fourth determining module is used to determine the pixel area of ​​the food ingredient based on the area ratio occupied by the food ingredient and the quantity of the food ingredient; The fifth determining module is used to determine the physical size of the food based on the conversion relationship between the pixel area and size of the food. The sixth determining module is used to determine the size classification of the food to be cleaned based on a preset size threshold and the physical size of the food.

[0149] In an optional embodiment, it further includes: The search key module is used to combine name identifiers and size hierarchy to form a composite search key; The first retrieval module is used to determine the first cleaning parameter as the initial cleaning parameter combination if a first cleaning parameter matching the composite retrieval key exists in the cleaning parameter knowledge base; or if a first cleaning parameter matching the composite retrieval key does not exist in the cleaning parameter knowledge base, the module determines the type of food to be cleaned, performs a downgraded retrieval based on the type of food, name identifier, and size classification, and matches the corresponding initial cleaning parameter combination in the cleaning parameter knowledge base.

[0150] In an optional embodiment, it further includes: The second retrieval module is used to determine the second cleaning parameter as the initial cleaning parameter combination if a second cleaning parameter matching the name identifier exists in the cleaning parameter knowledge base; or to perform a retrieval based on the type and size of the food ingredient if no second cleaning parameter exists in the cleaning parameter knowledge base. Also includes: The third retrieval module is used to determine the third cleaning parameter as the initial cleaning parameter combination if there is a third cleaning parameter in the cleaning parameter knowledge base that matches the type and size classification of the ingredients; or to retrieve the third cleaning parameter based on the type of ingredients if there is no third cleaning parameter in the cleaning parameter knowledge base. Also includes: The fourth retrieval module is used to determine the fourth cleaning parameter as the initial cleaning parameter combination if a fourth cleaning parameter matching the food type exists in the cleaning parameter knowledge base; or to obtain a preset safe default cleaning parameter combination as the initial cleaning parameter combination if no fourth cleaning parameter exists in the cleaning parameter knowledge base.

[0151] In an optional embodiment, it further includes: The seventh determination module is used to determine the first confidence level in the multidimensional recognition results; The fourth acquisition module is used to acquire the cleaning parameter knowledge base if the first confidence level is greater than or equal to the first preset threshold; or to send a result confirmation message if the first confidence level is less than the first preset threshold but greater than or equal to the second preset threshold; or to send an image retake prompt message if the first confidence level is less than the second preset threshold.

[0152] In an optional embodiment, it further includes: The fifth acquisition module is used to acquire the target physical control parameters in the initial cleaning parameter combination and the corresponding equipment limit parameters in the hardware capability parameter set; The parameter clamping module is used to clamp the target physical control parameter to the equipment limit parameter if the target physical control parameter is greater than the equipment limit parameter. The first instruction generation module is used to generate target cleaning instructions based on the equipment's limit parameters; the target cleaning instructions carry a degradation flag. The information sending module is used to send parameter downgrade prompts.

[0153] In an optional embodiment, it further includes: The sixth acquisition module is used to acquire the locally stored basic parameter table and determine the basic parameter table as the cleaning parameter knowledge base if the acquisition of the cleaning parameter knowledge base fails.

[0154] The apparatus and method embodiments in this application are based on the same application concept.

[0155] The methods and embodiments provided in this application can be executed on a computer terminal, server, or similar computing device. Taking running on a server as an example, Figure 6 This is a hardware structure block diagram of a server for a method of configuring food cleaning parameters provided in an embodiment of this application. Figure 6 As shown, the server 600 can vary significantly due to different configurations or performance. It may include one or more Central Processing Units (CPUs) 610 (CPUs 610 may include, but are not limited to, microprocessors (MCUs) or programmable logic devices (FPGAs), a memory 630 for storing data, and one or more storage media 620 (e.g., one or more mass storage devices) for storing application programs 623 or data 622. The memory 630 and storage media 620 may be temporary or persistent storage. The program stored in the storage media 620 may include one or more modules, each module may include a series of instruction operations on the server. Furthermore, the CPU 610 may be configured to communicate with the storage media 620 and execute the series of instruction operations stored in the storage media 620 on the server 600. Server 600 may also include one or more power supplies 660, one or more wired or wireless network interfaces 650, one or more input / output interfaces 640, and / or one or more operating systems 621, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0156] The input / output interface 640 can be used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of server 600. In one example, input / output interface 640 includes a network interface controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, input / output interface 640 may be a radio frequency (RF) module used for wireless communication with the Internet.

[0157] Those skilled in the art will understand that Figure 6 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, server 600 may also include... Figure 6 The more or fewer components shown, or having the same Figure 6 The different configurations shown.

[0158] This application provides an electronic device, which includes a processor and a memory. The memory stores at least one instruction, at least one program, code set, or instruction set. The processor loads and executes the at least one instruction, at least one program, code set, or instruction set to implement the above-described data processing method.

[0159] Embodiments of this application also provide a computer-readable storage medium, which can be disposed in a server to store at least one instruction, at least one program, code set, or instruction set related to implementing a food cleaning parameter configuration method in the method embodiment. The at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the above-mentioned food cleaning parameter configuration method.

[0160] Optionally, in this embodiment, the storage medium may be located at at least one of the multiple network servers in a computer network. Optionally, in this embodiment, the storage medium may include, but is not limited to, various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0161] As can be seen from the embodiments of the food cleaning parameter configuration method, apparatus, electronic device, or storage medium provided in this application, this application acquires image information of the food to be cleaned and a set of hardware capability parameters of the cleaning equipment; identifies the image information based on a preset multimodal model to obtain multidimensional identification results of the food to be cleaned; the multidimensional identification results include name identification and size classification; acquires a cleaning parameter knowledge base; constructs composite retrieval conditions based on name identification and size classification, and matches the corresponding initial cleaning parameter combination in the cleaning parameter knowledge base; verifies the initial cleaning parameter combination based on the set of hardware capability parameters, and generates a target cleaning instruction; the target cleaning instruction is used to drive the cleaning equipment to perform cleaning operations. In this application, by acquiring food images and equipment hardware capability parameters, using a cloud-based multimodal model to identify multidimensional food results, and constructing composite retrieval conditions to match initial cleaning parameters in the cleaning parameter knowledge base, and generating instructions to drive the cleaning equipment to execute after hardware capability verification, this application breaks through the limitations of traditional single-category identification, combines multidimensional food features to adaptively match the best cleaning parameters, improves cleaning efficiency and saves water resources, and ensures that parameters do not exceed limits through hardware capability verification, thus guaranteeing the safe operation of the equipment.

[0162] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims can be performed in a different order than that shown in the embodiments and still achieve the desired result. Additionally, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0163] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0164] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0165] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for configuring food cleaning parameters, characterized in that, include: Acquire image information of the food to be cleaned and a set of hardware capability parameters of the cleaning equipment; The image information is identified based on a preset multimodal model to obtain a multidimensional identification result of the food to be cleaned; The multidimensional recognition results include name identification and size classification; Access the cleaning parameter knowledge base; Based on the name identifier and the size classification, a composite search condition is constructed, and the corresponding initial cleaning parameter combination is matched in the cleaning parameter knowledge base; if the matching fails, the type of the food to be cleaned is determined, and a downgraded search is performed based on the food type, the name identifier and the size classification, and the corresponding initial cleaning parameter combination is matched in the cleaning parameter knowledge base. The initial cleaning parameter combination is verified based on the set of hardware capability parameters, and a target cleaning instruction is generated. The target cleaning command is used to drive the cleaning equipment to perform a cleaning operation; The step of recognizing the image information based on a preset multimodal model to obtain a multidimensional recognition result of the food to be cleaned includes: Based on the preset multimodal model, the image information is identified to obtain the visual reference objects in the image information and the name, type, quantity, and area ratio of the food to be cleaned. The size classification of the food to be cleaned is determined based on the reference physical size of the visual reference object, the area ratio of the reference object to the visual reference object, the quantity of the food, the area occupied by the food, and a preset size threshold. If there is only one type of ingredient, obtain the name identifier and size classification of the ingredient to be cleaned; or if there are multiple types of ingredients, determine the main ingredient among the multiple ingredients to be cleaned based on the size classification, determine the size classification of the main ingredient as the size classification of the ingredient to be cleaned, and determine the name identifier of the main ingredient as the name identifier of the ingredient to be cleaned; the size classification of the main ingredient is greater than the size classification of other ingredients among the multiple ingredients to be cleaned besides the main ingredient; The multidimensional identification result is determined based on the name and size classification of the food items to be cleaned.

2. The method for configuring food cleaning parameters according to claim 1, characterized in that, The determination of the size classification of the food to be cleaned based on the reference physical size of the visual reference object, the area ratio of the reference object, the quantity of the food, the area occupied by the food, and a preset size threshold includes: Obtain the reference physical size and area ratio of the visual reference object; The size conversion relationship between physical size and pixel area is determined based on the reference physical size and the area ratio occupied by the reference object; The pixel area of ​​the food ingredient is determined based on the area ratio occupied by the food ingredient and the quantity of the food ingredient; The physical dimensions of the food are determined based on the pixel area of ​​the food and the size conversion relationship. The size classification of the food to be cleaned is determined based on a preset size threshold and the physical size of the food.

3. The method for configuring food cleaning parameters according to claim 1, characterized in that, The step of constructing composite search conditions based on the name identifier and the size classification, and matching the corresponding initial cleaning parameter combination in the cleaning parameter knowledge base, includes: The name identifier and the size hierarchy are combined to form a composite search key; If a first cleaning parameter matching the composite search key exists in the cleaning parameter knowledge base, the first cleaning parameter is determined as the initial cleaning parameter combination; or if a first cleaning parameter matching the composite search key does not exist in the cleaning parameter knowledge base, the type of the food to be cleaned is determined, and a downgraded search is performed based on the type of food, the name identifier, and the size classification to match the corresponding initial cleaning parameter combination in the cleaning parameter knowledge base.

4. The method for configuring food cleaning parameters according to claim 3, characterized in that, The step of performing a downgraded search based on the food type, name identifier, and size classification, and matching the corresponding initial cleaning parameter combination in the cleaning parameter knowledge base, includes: If a second cleaning parameter matching the name identifier exists in the cleaning parameter knowledge base, the second cleaning parameter is determined as the initial cleaning parameter combination; or if the second cleaning parameter does not exist in the cleaning parameter knowledge base, a search is performed based on the food type and the size classification. The retrieval based on the food type and size classification includes: If a third cleaning parameter that matches the food type and size classification exists in the cleaning parameter knowledge base, the third cleaning parameter is determined as the initial cleaning parameter combination; or if the third cleaning parameter does not exist in the cleaning parameter knowledge base, a search is performed based on the food type. The search based on the type of food ingredient includes: If a fourth cleaning parameter matching the type of food exists in the cleaning parameter knowledge base, the fourth cleaning parameter is determined as the initial cleaning parameter combination; or if the fourth cleaning parameter does not exist in the cleaning parameter knowledge base, a preset safe default cleaning parameter combination is obtained as the initial cleaning parameter combination.

5. The method for configuring food cleaning parameters according to claim 1, characterized in that, The multidimensional recognition result also includes confidence level; the acquisition of the cleaning parameter knowledge base includes: Determine a first confidence level from the multidimensional recognition results; If the first confidence level is greater than or equal to the first preset threshold, obtain the cleaning parameter knowledge base; or if the first confidence level is less than the first preset threshold but greater than or equal to the second preset threshold, send a result confirmation message; or if the first confidence level is less than the second preset threshold, send an image retake prompt message.

6. The method for configuring food cleaning parameters according to claim 1, characterized in that, The step of verifying the initial cleaning parameter combination based on the hardware capability parameter set and generating a target cleaning instruction includes: Obtain the target physical control parameters in the initial cleaning parameter combination and the corresponding equipment limit parameters in the hardware capability parameter set; If the target physical control parameter is greater than the device limit parameter, then the target physical control parameter is clamped to the device limit parameter; The target cleaning instruction is generated based on the device's limit parameters; the target cleaning instruction carries a degradation flag. Send a parameter downgrade prompt message.

7. The method for configuring food cleaning parameters according to claim 1, characterized in that, After obtaining the cleaning parameter knowledge base, the following is also included: If the cleaning parameter knowledge base fails to be retrieved, a locally stored basic parameter table is retrieved and the basic parameter table is identified as the cleaning parameter knowledge base.

8. A device for configuring food cleaning parameters, characterized in that, The device includes: The first acquisition module is used to acquire image information of the food to be cleaned and a set of hardware capability parameters of the cleaning equipment. The image recognition module is used to recognize the image information based on a preset multimodal model to obtain a multidimensional recognition result of the food to be cleaned; the multidimensional recognition result includes name identification and size classification; The second acquisition module is used to acquire the cleaning parameter knowledge base; The retrieval and matching module is used to construct composite retrieval conditions based on the name identifier and the size classification, and match the corresponding initial cleaning parameter combination in the cleaning parameter knowledge base; if the matching fails, the type of the food to be cleaned is determined, and a downgraded retrieval is performed based on the food type, the name identifier and the size classification, and the corresponding initial cleaning parameter combination is matched in the cleaning parameter knowledge base. The instruction generation module is used to verify the initial cleaning parameter combination based on the hardware capability parameter set and generate a target cleaning instruction; the target cleaning instruction is used to drive the cleaning equipment to perform a cleaning operation. The step of recognizing the image information based on a preset multimodal model to obtain a multidimensional recognition result of the food to be cleaned includes: Based on the preset multimodal model, the image information is identified to obtain the visual reference objects in the image information and the name, type, quantity, and area ratio of the food to be cleaned. The size classification of the food to be cleaned is determined based on the reference physical size of the visual reference object, the area ratio of the reference object to the visual reference object, the quantity of the food, the area occupied by the food, and a preset size threshold. If there is only one type of ingredient, obtain the name identifier and size classification of the ingredient to be cleaned; or if there are multiple types of ingredients, determine the main ingredient among the multiple ingredients to be cleaned based on the size classification, determine the size classification of the main ingredient as the size classification of the ingredient to be cleaned, and determine the name identifier of the main ingredient as the name identifier of the ingredient to be cleaned; the size classification of the main ingredient is greater than the size classification of other ingredients among the multiple ingredients to be cleaned besides the main ingredient; The multidimensional identification result is determined based on the name and size classification of the food items to be cleaned.

9. An electronic device, characterized in that, The electronic device includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, code set, or instruction set, and the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the food cleaning parameter configuration method as described in any one of claims 1-7.

Images