Washing strategy generation method and system based on visual recognition, and storage medium

Abstract

This invention relates to a method, system, and storage medium for generating washing strategies based on visual recognition. The method includes the following steps: acquiring a first image set of clothes to be washed; inputting the first image set into a pre-stored clothing management database within the clothing care device for matching to obtain a first clothing information set; acquiring a second image set of clothes to be washed; inputting the second image set into a pre-trained local washing inference model for inference to obtain a second clothing information set; inputting the second image set and the second clothing information set into the first clothing information set for retrieval to match the first clothing material information and preset washing strategy information corresponding to each individual piece of clothing to be washed in the second image set. This effectively avoids problems such as incomplete washing or over-washing caused by human judgment errors, thereby protecting the texture and color of the clothing and avoiding problems such as over-washing or detergent residue.

Description

Technical Field

[0001] This invention relates to the field of washing machine technology, and specifically to a method, system, and storage medium for generating washing strategies based on visual recognition. Background Technology

[0002] In the commercial laundry and care industry, traditional washing strategies rely primarily on user judgment. Users must manually select the appropriate washing program and detergent dosage based on their personal experience and observation, considering factors such as the fabric and type of clothing, as well as the type and location of stains. However, this approach often has significant limitations.

[0003] On the one hand, because users may lack professional washing knowledge, they often struggle to accurately determine the type of stain and the material of the clothing, which may lead to inappropriate washing procedures and inaccurate detergent dosage, thus affecting the washing effect. On the other hand, the process of users making their own judgments is cumbersome and time-consuming, reducing washing efficiency, especially when there are many clothes. In addition, traditional washing strategies are prone to missing stained areas, especially those hidden in the folds or overlapping parts of clothing. These hard-to-detect stains often result in incomplete washing, affecting the overall washing effect of the clothes.

[0004] This application relates to the development of a visual recognition-based washing strategy generation method, system, and storage medium to solve the aforementioned problems. Summary of the Invention

[0005] To achieve the above-mentioned objectives and other advantages of the present invention, a first objective of the present invention is to provide a method for generating a washing strategy based on visual recognition, comprising the following steps: Acquire a first image set of clothes to be washed; wherein the first image set is an overall image of multiple stacked clothes acquired by an image acquisition device deployed in the garment care device; The first image set is input into the clothing management database pre-stored in the clothing care device for matching to obtain a first clothing information set; wherein, the first clothing information set includes at least the washing interaction information corresponding to each individual garment to be washed; wherein, the washing interaction information includes at least the overall image corresponding to each individual garment to be washed and its corresponding first clothing material information and preset washing strategy information. Acquire a second image set of the garments to be washed; wherein the second image set consists of detailed images of the stained areas of each individual garment acquired by an image acquisition device deployed within the garment care device; The second image set is input into a pre-trained local washing inference model for inference to obtain a second clothing information set; The second image set and the second clothing information set are input into the first clothing information set for retrieval, so as to match the first clothing material information and preset washing strategy information corresponding to each piece of clothing to be washed in the second image set.

[0006] Furthermore, the first clothing information set includes the total number of the clothing items to be washed.

[0007] Furthermore, the second clothing information set includes clothing parameter information, stain data information, and suggested washing strategy information corresponding to the individual garment.

[0008] Furthermore, the clothing parameter information includes second clothing material information corresponding to the detailed image of the stained area of ​​the single garment; the stain data information includes at least one or more of the following: clothing stain type, clothing stain area, and clothing stain intensity.

[0009] Furthermore, it also includes the following steps: Determine whether the suggested washing strategy information matches the preset washing strategy information; If the suggested washing strategy information matches the preset washing strategy information, then the suggested washing strategy information is executed; If the suggested washing strategy information does not match the preset washing strategy information, the preset washing strategy information is adjusted using the stain data information to generate the washing strategy information to be executed.

[0010] Furthermore, it also includes the following steps: The second clothing information set is transmitted back to the user through interaction; Determine whether the second clothing information set needs to be modified; If correction is required, the user-corrected second clothing information set is input into the first clothing information set for retrieval; If no correction is required, the second clothing information set is input into the first clothing information set for retrieval.

[0011] Furthermore, it also includes the following steps: Determine whether it is necessary to combine cloud servers for reasoning; If it is necessary to combine the cloud server for inference, the second image set is input into the remote washing inference model deployed in the cloud server for inference to obtain the second clothing information set; If inference is not required using the cloud server, the second image set is input into a pre-trained local washing inference model to obtain the second clothing information set.

[0012] Furthermore, when inference is required in conjunction with the cloud server, the following steps are also included: The second set of clothing information is fed back to the local washing inference model to optimize the local washing inference model.

[0013] Further, the first image set is input into a clothing management database pre-stored in the clothing care device for matching to obtain a first clothing information set, including the following steps: Multiple clothing regions of the overall image of the stacked clothing are obtained as the target clothing image set for washing; The individual images from the set of images of the target clothes to be washed are sequentially input into the clothes management database; The clothing features of the single image are compared one by one with the images in the clothing management database; Determine if the match was successful; If the match is successful, retrieve the first clothing information set; If the matching fails, the single image is input into the pre-trained local washing inference model for inference to obtain the first clothing information set; The first set of clothing information and the single image are stored in the clothing management database.

[0014] Furthermore, the clothing features correspond to the information set classification stored in the clothing management database, and include at least one or more of clothing color, clothing type, and clothing size.

[0015] Further, the second image set and the second clothing information set are input into the first clothing information set for retrieval, in order to match the first clothing material information and preset washing strategy information corresponding to each individual piece of clothing to be washed in the second image set, including the following steps: Obtain an overall image of a single garment from the first garment information set; Acquire a multi-target region image of the entire image of the single garment; wherein, the multi-target region image consists of multiple images containing stained areas; Obtain the multi-target region image and the local feature map in the second image set; Local feature maps in the second image set are retrieved one by one from the local feature images of the multi-target region image to obtain candidate target images; Retrieve the first clothing material information of the candidate target image; The second clothing material information is input into the first clothing material information for retrieval, so as to match the first clothing material information corresponding to the second clothing information; Obtain the preset washing strategy information corresponding to the first clothing material information.

[0016] A second objective of this invention is to provide a washing strategy generation system based on visual recognition, comprising the following modules: An image acquisition module is configured to acquire a first image set of clothes to be washed; wherein the first image set is an overall image of multiple stacked clothes acquired by an image acquisition device deployed within the garment care device; The image acquisition module is also used to acquire a second image set of the clothes to be washed; wherein, the second image set is a detailed image of the stained area of ​​each individual garment acquired by the image acquisition device deployed in the garment care device; The information acquisition module is configured to input the first image set into a clothing management database pre-stored in the clothing care device for matching, so as to obtain a first clothing information set; wherein, the first clothing information set includes at least washing interaction information corresponding to each individual garment to be washed; the washing interaction information includes at least the overall image corresponding to each individual garment to be washed and its corresponding first clothing material information and preset washing strategy information. The model inference module is configured to input the second image set into a pre-trained local washing inference model for inference to obtain a second clothing information set; The washing strategy generation module is configured to input the second image set and the second clothing information set into the first clothing information set for retrieval, so as to match the first clothing material information and preset washing strategy information corresponding to each piece of clothing to be washed in the second image set.

[0017] The cloud server module is configured to deploy a pre-trained remote washing inference model.

[0018] Furthermore, the image acquisition module includes a wide-angle camera unit and a macro camera unit; wherein, the wide-angle camera unit is used to acquire a first image set of the clothes to be washed, and the macro camera unit is used to acquire a second image set of the clothes to be washed.

[0019] A third objective of the present invention is to provide a readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements a washing strategy generation method based on visual recognition.

[0020] A fourth objective of this invention is to provide an electronic device comprising: a processor and a memory, the memory being used to store one or more programs; when the one or more programs are executed by the processor, a method for generating a washing strategy based on visual recognition is implemented.

[0021] Compared with the prior art, the beneficial effects of the present invention are: This invention relates to a method, system, and storage medium for generating washing strategies based on visual recognition. By acquiring an overall image of a stack of multiple garments to be washed and detailed images of the stained areas of each individual garment, and combining this with a pre-stored garment management database within a garment care device and a pre-trained local washing inference model, this invention can accurately identify the garment material, stain type, and location. It then generates a washing strategy based on the specific condition of each garment, providing a personalized washing service. This effectively avoids problems such as incomplete or over-washing due to human error, thereby protecting the fabric texture and color of the garments and preventing over-washing or detergent residue.

[0022] This invention, through automated and intelligent image recognition and reasoning processes, can rapidly generate washing strategies, significantly shortening washing preparation time and improving washing efficiency. For the commercial laundry and care industry, this means the ability to handle more laundry, enhancing service capabilities and customer satisfaction. Furthermore, with increased usage time, the pre-trained model and laundry management database of this invention can continuously learn and accumulate data, further optimizing the washing strategy generation process. This allows for accurate identification of clothing materials, stain types, and locations, quickly generating precise washing strategies and contributing to improved consistency and stability of washing results.

[0023] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it according to the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Specific embodiments of the present invention are given in detail below with reference to the accompanying drawings. Attached Figure Description

[0024] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 This is the flow chart of the vision recognition-based washing strategy generation method of this application. Figure 1 ; Figure 2 This is a flowchart of the process for obtaining the first clothing information set as described in Example 1; Figure 3 This is a schematic diagram of the washing reasoning model described in Example 1. Figure 1 ; Figure 4 This is a schematic diagram of the washing reasoning model described in Example 1. Figure 2 ; Figure 5 The flow chart of the vision recognition-based washing strategy generation method in Example 1 Figure 2 ; Figure 6The flow chart of the vision recognition-based washing strategy generation method in Example 1 Figure 3 ; Figure 7 This is a flowchart illustrating the process of matching the first garment material information and preset washing strategy information corresponding to each individual garment to be washed in the second image set, as described in Example 1. Figure 8(a) is a local feature map of the multi-target region image described in Example 1. Figure 1 ; Figure 8(b) shows the candidate target image described in Example 1. Figure 1 ; Figure 9(a) is a local feature map of the multi-target region image described in Example 1. Figure 2 ; Figure 9(b) shows the candidate target image described in Example 1. Figure 2 ; Figure 10(a) is a local feature map of the multi-target region image described in Example 1. Figure 3 ; Figure 10(b) shows the candidate target image described in Example 1. Figure 3 ; Figure 11 The flowchart of the vision recognition-based washing strategy generation method in the embodiment is as follows. Figure 4 ; Figure 12 This is a schematic diagram of the vision recognition-based washing strategy generation system in Example 2; Figure 13 This is a schematic diagram of a computer-readable storage medium in Example 3; Figure 14 This is a schematic diagram of the electronic device in Example 4. Detailed Implementation

[0025] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments. It should be noted that, without conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0026] In the following description, the use of suffixes such as "module," "part," or "unit" to denote elements is solely for the purpose of illustrative purposes and has no specific meaning in itself. Therefore, "module," "part," or "unit" may be used interchangeably.

[0027] Example 1 This invention provides a method for generating washing strategies based on visual recognition, such as... Figure 1 As shown, the specific steps include the following: S101, acquire a first image set of clothes to be washed; wherein, the first image set is an overall image of multiple stacked clothes acquired by an image acquisition device deployed in the clothes care device; S102, the first image set is input into the clothing management database pre-stored in the clothing care device for matching to obtain the first clothing information set; wherein, the first clothing information set includes at least the washing interaction information corresponding to each individual piece of clothing to be washed; the washing interaction information includes at least the overall image corresponding to each individual piece of clothing to be washed and its corresponding first clothing material information and preset washing strategy information. S103, acquire a second image set of the clothes to be washed; wherein, the second image set is a detailed image of the stained area of ​​each individual garment acquired by an image acquisition device deployed in the garment care device; S104, The second image set is input into the pre-trained local washing inference model for inference to obtain the second clothing information set; S105, input the second image set and the second clothing information set into the first clothing information set for retrieval, so as to match the first clothing material information and preset washing strategy information corresponding to each piece of clothing to be washed in the second image set.

[0028] In some embodiments, the image acquisition device described in step S101 is a vision module.

[0029] In a preferred embodiment, the vision module includes a wide-angle camera unit, a macro camera unit, and a processor; wherein the wide-angle camera unit is used to acquire a first set of images of the clothes to be washed, and the macro camera unit is used to acquire a second set of images of the clothes to be washed.

[0030] For example, when the garment care device is activated, the visual recognition module uses a wide-angle camera unit to capture an overall image of the stacked state of the multiple garments to be washed, forming a first image set. It should be understood that the first image set includes multiple images of the stacked garments taken from different angles to ensure that each garment to be washed is captured with at least one appearance image.

[0031] In some embodiments, the first clothing information set in step S102 may also include the total number of the clothing to be washed.

[0032] In some embodiments, the second clothing information set in step S104 includes clothing parameter information, stain data information, and suggested washing strategy information corresponding to a single garment.

[0033] In a preferred embodiment, the clothing parameter information includes second clothing material information corresponding to a detailed image of the stained area of ​​the single garment; the stain data information includes at least one or more of clothing stain type, clothing stain area, and clothing stain intensity.

[0034] In some embodiments, step S102, which involves inputting the first image set into a pre-existing clothing management database within the clothing care device for matching to obtain a first clothing information set, is as follows: Figure 2 As shown, it includes the following steps: S1021, acquire multiple clothing regions of the overall image of the stack of multiple clothes as a set of images of the target clothes for washing; S1022, input the individual images from the set of images of the target clothes to the clothes management database in sequence; S1023, compare the clothing features of the single image with the clothing features of the images in the clothing management database one by one; S1024, determine whether the clothing features of the single image match the clothing features of the images in the clothing management database; S1025a, If the match is successful, retrieve the first clothing information set; S1025b, if the matching fails, the single image is input into the pre-trained local washing inference model for inference to obtain the first clothing information set; S1026, the first clothing information set and the single image are stored in the clothing management database.

[0035] In a preferred embodiment, the washing reasoning model described in step S1025b is as follows: Figure 3 As shown, it includes the following units: Clothing feature extraction unit: acquires the single image, performs convolution processing on the single image through multiple convolutional layers and variable convolutional layers to extract clothing features from the single image; performs fast spatial pyramid pooling processing on the clothing features to improve the robustness of the clothing features and the generalization ability of the model.

[0036] Multi-scale feature fusion unit: The clothing features are fused by splicing and upsampling to obtain comprehensive features with information at different scales; the comprehensive features are then processed by convolution and variable convolution layers to enhance the features.

[0037] Regression prediction unit: The fused and enhanced features are used for regression prediction, that is, the clothing is identified and detected based on the extracted features to output the first clothing information set.

[0038] In a preferred embodiment, the clothing features correspond to the information set classification stored in the clothing management database, including at least one or more of clothing color, clothing type, and clothing size.

[0039] In another preferred embodiment, in order to extract and analyze clothing information more accurately, different inference networks can be used to extract different types of clothing features from different input clothing information. The different inference networks have similar network structures, but their parameter settings, training data and input / output interfaces are different for the feature types they process, so as to extract various feature information of clothing more accurately and efficiently.

[0040] Specifically, it should be understood that, for example, when it is necessary to extract the color features of clothing, the color features are specifically optimized in the color recognition network, and when the image data of clothing is input, the output is the recognized color category or color value of clothing.

[0041] When it is necessary to extract clothing type features, the network structure is adjusted so that when the input is clothing image data, the output is the identified clothing type.

[0042] When it is necessary to extract stain features, the location and type information of the stain are output when the image data of the clothing is input.

[0043] The washing inference model involved in this application is specifically a convolutional neural network model in deep learning. Through feature extraction of multiple convolutional layers and variable convolutional layers, as well as the application of techniques such as fast spatial pyramid pooling and multi-scale feature fusion, it efficiently processes clothing images and accurately identifies and detects the first clothing information set of the target clothing, which has broad application prospects.

[0044] In a preferred embodiment, the washing reasoning model involved in this application can also be used for strategy selection, such as... Figure 4 As shown, this is information on washing strategies used to optimize the washing process.

[0045] Specifically, it should be understood that the following components are included: Value function model components: include a Q-network and a target Q-network, wherein the Q-network is used to predict the Q-value (i.e., expected return) of taking each action in the current state, and the target Q-network is used to generate the target Q-value required to update the parameters of the Q-network.

[0046] The deep Q-learning network component includes Q-prediction, a loss function, a Q-target, and a learning module. Specifically, the Q-prediction module selects the action with the highest Q-value based on the current state and the Q-network's prediction; the loss function module measures the difference between the Q-value predicted by the Q-network and the target Q-value, thereby guiding the update of network parameters; the Q-target module generates a target Q-value for updating the Q-network based on environmental feedback and actions already taken; and the learning module updates the Q-network parameters based on the results of the loss function calculation to improve prediction accuracy.

[0047] Environmental feedback and sensor components: Information from the environment (such as washing effect and first and second garment information sets) is received through sensors and used as feedback to update the Q network parameters, thereby optimizing the next washing strategy.

[0048] Experience replay mechanism component: Used to store the experiences generated by the model during its interaction with the environment. These experiences are sequences of states, actions, rewards, and the next state. By replaying these experiences, the model can learn more effectively and improve learning efficiency.

[0049] Network parameter update component: Used to periodically update the parameters of the Q network to ensure that it can adapt to the ever-changing environment and maintain the accuracy and stability of predictions.

[0050] In a preferred embodiment, a greedy algorithm can be used for selection in Q-prediction. Its core lies in exploring using probability ε to obtain the optimal policy information. Specifically, it should be understood that the magnitude of ε directly affects the learning efficiency and performance of the agent. A smaller ε value leads to more exploitation and less exploration, potentially causing premature convergence to a local optimum; a larger ε value increases exploration, helping to discover better global policies, but may also slow down the learning process. The ε value can be adjusted according to the specific application scenario to improve learning efficiency and stability.

[0051] The washing reasoning model involved in this application achieves intelligent decision-making during the washing process by continuously interacting with the environment, learning, and optimizing the decision-making process to generate better washing strategy information. Through the combination of a value function model and a deep Q-learning network, the model can accurately predict the expected returns of different actions in the current state and select the optimal action. Simultaneously, environmental feedback and experience replay mechanisms further improve the model's learning efficiency and generalization ability, while increasing washing efficiency and quality and reducing energy consumption and costs.

[0052] This application acquires an overall image of multiple stacked garments using a wide-angle camera unit, and segments the image into multiple garment regions using image segmentation technology. Specifically, these multiple garment regions include the individual garment regions of all stacked garments to be washed, thus forming a target garment image set for washing. Subsequently, the images in the target garment image set are sequentially input into a garment management database. The database compares the extracted garment images with the images in the garment management database one by one based on their features (such as color, shape, texture, type, etc.). Based on the comparison results, it determines whether the input garment image matches a certain image in the garment management database. If the match is successful, the corresponding first garment information set is retrieved from the garment management database. If the match is unsuccessful, the image is input into a pre-trained local washing inference model, which infers information such as the material and stain type of the garment, thereby generating the first garment information set. Therefore, this application can quickly identify whether the clothes to be washed are in the database using automated image recognition with reduced manual intervention, thereby effectively retrieving the corresponding first set of clothing information. If the clothes are not in the database, the clothing information set can still be obtained by using a pre-trained local washing inference model, enhancing the system's flexibility and adaptability. In addition, the newly identified clothing information set and images are stored in the clothing management database, continuously enriching the database content and helping to improve the accuracy and efficiency of future matching, forming a virtuous cycle.

[0053] In some embodiments, before step S105, such as Figure 5 As shown, it also includes the following steps: S131, the second clothing information is collected and transmitted back to the user through interaction; S132, determine whether the second clothing information set needs to be modified; S133a, If correction is required, the user-corrected second clothing information set is input into the first clothing information set for retrieval; S133b, if no correction is needed, the second clothing information set is input into the first clothing information set for retrieval.

[0054] During image recognition, factors such as lighting conditions and occlusion can affect the quality of image recognition. This application addresses this by transmitting a second clothing information set back to the user, allowing for user interaction to verify and correct the clothing parameter information and stain data in the recognized second clothing information set. This helps ensure the accuracy of the recognized information. Furthermore, allowing users to correct the automatic recognition results increases user trust in the system, thereby improving the user experience.

[0055] In some embodiments, before step S105, such as Figure 6 As shown, it also includes the following steps: S141, Determine whether reasoning needs to be performed in conjunction with the cloud server; S142a, If it is necessary to combine the cloud server for inference, the second image set is input into the remote washing inference model deployed in the cloud server for inference to obtain the second clothing information set; S142b, If inference is not required in conjunction with the cloud server, the second image set is input into the pre-trained local washing inference model for inference to obtain the second clothing information set.

[0056] In a preferred embodiment, after step S142a, the following step is further included: The second set of clothing information is fed back to the local washing inference model to optimize the local washing inference model.

[0057] After the second set of clothing information is transmitted back, the user can determine whether to combine it with the cloud server for inference based on the received information. When cloud inference is required, the second image set is input into the remote washing inference model on the cloud server for inference. Because the cloud server has more powerful computing capabilities and richer data resources, it can more accurately obtain the second set of clothing information, thereby improving the accuracy and efficiency of washing strategy generation. Furthermore, the cloud server enables centralized management and sharing of data, allowing different devices and locations to access the latest data and models. This means that the washing strategy generation method can be updated and optimized at any time based on the latest data and algorithms to adapt to different types of clothing and stain conditions.

[0058] In some embodiments, the suggested washing strategy information, preset washing strategy information, and specific washing strategy of the present invention include at least washing process selection, washing parameter selection, detergent selection, and detergent dosing control.

[0059] In some embodiments, step S105 involves inputting the second image set and the second clothing information set into the first clothing information set for retrieval, in order to match the first clothing material information and preset washing strategy information corresponding to each individual piece of clothing to be washed in the second image set, such as... Figure 7 As shown, it includes the following steps: S1051, Obtain an overall image of a single garment from the first garment information set; S1052, acquire a multi-target region image of the overall image of the single garment; wherein, the multi-target region image is multiple images containing the stained area; S1053, acquire the multi-target region image and the local feature map in the second image set; S1054, retrieve the local feature maps in the second image set one by one from the local feature images of the multi-target region image to obtain candidate target images; S1055, retrieve the first clothing material information of the candidate target image; S1056, Input the second clothing material information into the first clothing material information for retrieval, so as to match the first clothing material information corresponding to the second clothing information; S1057, Obtain the preset washing strategy information corresponding to the first clothing material information.

[0060] Example 1: The local feature image of the acquired multi-target region image is shown in Figure 8(a), and the candidate target image is shown in Figure 8(b). Specifically, the first clothing material information of the local feature image acquired in Figure 8(a) is cotton-linen blend, the stain type is bloodstain, and the acquired preset washing strategy information is as follows: Set the washing mode to gentle wash and the water temperature to 30℃ to avoid the bloodstains from coagulating and damaging the cotton blend fibers. Set the detergent to contain bio-enzymes to effectively break down the proteins in the bloodstains and enhance the stain removal effect.

[0061] Example 2: The local feature image of the acquired multi-target region image is shown in Figure 9(a), and the candidate target image is shown in Figure 9(b). Specifically, the first clothing material information of the local feature image acquired in Figure 9(a) is silk, the stain type is coffee stain, and the acquired preset washing strategy information is as follows: Set the washing mode to silk fine wash, the water temperature to 25℃, and the detergent to neutral or weakly acidic detergent to effectively dissolve stains without damaging the silk fiber structure.

[0062] Example 3: The local feature image of the acquired multi-target region image is shown in Figure 10(a), and the candidate target image is shown in Figure 10(b). Specifically, the first clothing material information of the local feature image acquired in Figure 10(a) is a blend of cotton and ice silk, the stain type is fruit juice stain, and the acquired preset washing strategy information is as follows: Set the washing mode to mixed fabric wash and the water temperature to 28℃ to avoid shrinkage of clothes or deformation of ice silk fibers due to excessively high water temperature. Excessively low water temperature will affect the dissolution and stain removal effect of the detergent. Set the detergent to contain neutral or weakly acidic detergent and add an appropriate amount of white vinegar to the detergent to enhance the stain removal effect.

[0063] In some embodiments, after step S105, as Figure 11 As shown, it also includes the following steps: S106, determine whether the suggested washing strategy information matches the preset washing strategy information; S107a, If the suggested washing strategy information matches the preset washing strategy information, then the suggested washing strategy information is executed; S107b, if the suggested washing strategy information does not match the preset washing strategy information, the preset washing strategy information is adjusted using the stain data information to generate the washing strategy information to be executed.

[0064] The visual recognition-based washing strategy generation method disclosed in this application enables comprehensive analysis of laundry and the formulation of personalized washing strategies. First, a first image set of the laundry to be washed—an overall image of multiple stacked garments—is acquired, providing a rich data foundation for subsequent analysis. Next, using a pre-stored clothing management database within the clothing care device, the overall image of the stacked garments is matched to quickly retrieve the first clothing information set, namely the total number of laundry items to be washed and the washing interaction information corresponding to each individual item. Subsequently, detailed images of the stained areas of each individual item of laundry are input into a pre-trained local washing inference model for reasoning to obtain a second clothing information set. This process combines advanced image recognition and deep learning technologies, enabling accurate identification of stain types and degrees, thereby providing more personalized washing suggestions. Finally, the second image set and the second clothing information set are input into the first clothing information set for retrieval to match the first clothing material information and preset washing strategy information corresponding to each individual item of laundry in the second image set. By comparing the suggested washing strategy information with the preset washing strategy information, the method can determine whether the two match and make adjustments according to the actual situation to generate the final washing strategy information.

[0065] The method described in this application also incorporates a user interaction and feedback mechanism, allowing users to correct the second clothing information set through an interactive interface, thereby improving the system's accuracy and user satisfaction. Furthermore, user feedback data can be used to further optimize the local washing inference model, creating a virtuous cycle and continuously improving system performance.

[0066] The method described in this application also supports inference and model updates in conjunction with cloud servers. When more complex analysis or model optimization is required, relevant data can be uploaded to the cloud server, inference can be performed using a remote washing inference model, and the results can be sent back to the local system. This process not only enhances the scalability and adaptability of the system but also ensures that the system can keep pace with technological developments and maintain its competitiveness and practicality.

[0067] Example 2 This invention provides a washing strategy generation system based on visual recognition, such as... Figure 12 As shown, it includes the following modules: An image acquisition module is configured to acquire a first image set of clothes to be washed; wherein the first image set is an overall image of multiple stacked clothes acquired by an image acquisition device deployed within the garment care device; The image acquisition module is also used to acquire a second image set of the clothes to be washed; wherein, the second image set is a detailed image of the stained area of ​​each individual garment acquired by the image acquisition device deployed in the garment care device; The information acquisition module is configured to input the first image set into a clothing management database pre-stored in the clothing care device for matching, so as to obtain a first clothing information set; wherein, the first clothing information set includes the total number of the clothes to be washed and the washing interaction information corresponding to each individual piece of clothing to be washed; the washing interaction information includes at least the overall image of each individual piece of clothing to be washed and its corresponding first clothing material information, and preset washing strategy information. The model inference module is configured to input the second image set into a pre-trained local washing inference model for inference to obtain a second clothing information set; The washing strategy generation module is configured to input the second image set and the second clothing information set into the first clothing information set for retrieval, so as to match the first clothing material information and preset washing strategy information corresponding to each piece of clothing to be washed in the second image set.

[0068] The cloud server module is configured to deploy a pre-trained remote washing inference model.

[0069] In some embodiments, the image acquisition module includes a wide-angle camera unit and a macro camera unit; wherein the wide-angle camera unit is used to acquire a first set of images of the clothes to be washed, and the macro camera unit is used to acquire a second set of images of the clothes to be washed.

[0070] In some embodiments, the system further includes the following modules: The image feedback module is used to feed back the second clothing information set through user interaction; The data correction module is used to determine whether the second clothing information set needs to be corrected after user interaction; If correction is required, the user-corrected second clothing information set is input into the first clothing information set for retrieval; If no correction is required, the second clothing information set is input into the first clothing information set for retrieval.

[0071] The model update module is used to determine whether the local washing inference model needs to be updated; When the local washing inference model needs to be updated, the second clothing information set is sent back to the local washing inference model to optimize the local washing inference model.

[0072] The visual recognition-based washing strategy generation system of this invention integrates multiple core modules, including image acquisition, information gathering, model inference, washing strategy generation, and a cloud server, through a highly integrated design, providing users with an intelligent and automated washing strategy generation system. First, the image acquisition module uses a wide-angle camera and a macro camera to capture overall images of stacked garments and detailed images of individual garment stain areas. Next, the information gathering module inputs the stacked garment images into a pre-stored garment management database for matching, quickly retrieving key data such as the total number of garments, overall images of individual garments, corresponding first garment material information, and preset washing strategy information, reducing the time spent on manual querying and judgment. Then, the model inference module uses a pre-trained local washing inference model to intelligently analyze the images of stain areas on individual garments, extracting stain data and suggested washing strategy information, achieving accurate identification of stain type, severity, and garment material, providing a precise basis for washing strategy formulation. Finally, the washing strategy generation module automatically adjusts the washing strategy based on the matching between the suggested washing strategy information and the preset washing strategy information, generating specific washing strategy information. If the two match, the suggested washing strategy is executed directly; otherwise, the preset washing strategy is adjusted based on stain data to meet the personalized washing needs of different materials and stains. Furthermore, the system supports user interaction and data correction. The image feedback module and data correction module allow users to view and correct the clothing material and stain data identified by the system through an interactive interface, improving system flexibility and user satisfaction. The corrected data can be re-input into the washing inference model for further optimization of the system's recognition capabilities and accuracy. Simultaneously, the model update module can determine whether the local washing inference model needs updating and, based on actual needs, feeds new data back to the model, enabling continuous model optimization and upgrades. This step ensures the system can continuously adapt to new washing requirements and stain types, improving its practicality and competitiveness. Finally, the deployment of a cloud server provides powerful computing support and data synchronization capabilities. The cloud server module deploys a pre-trained remote washing inference model, providing backup and expansion possibilities for the system, ensuring stable operation and real-time data synchronization.

[0073] Example 3 This invention also provides a computer-readable storage medium, such as... Figure 13As shown, it stores program instructions, which, when executed, implement the washing strategy generation method based on visual recognition as described in Embodiment 1 above.

[0074] The program instructions are stored in a computer-readable storage medium (such as a CD-ROM, USB flash drive, or portable hard drive) or on a network, and include several computer program instructions to cause a computing device (such as a personal computer, server, or network device) to execute the above-described method according to the embodiments of this application.

[0075] Example 4 This invention also provides an electronic device, such as... Figure 14 As shown, it includes: a processor and a memory, the memory being used to store one or more programs; when the one or more programs are executed by the processor, the visual recognition-based washing strategy generation method described in Embodiment 1 above is implemented.

[0076] Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention, and other modifications can be easily made by those skilled in the art. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and examples shown and described herein.

[0077] The apparatus, electronic device, and non-volatile computer storage medium and method provided in the embodiments of this specification are corresponding. Therefore, the apparatus, electronic device, and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding apparatus, electronic device, and non-volatile computer storage medium will not be repeated here.

[0078] Those skilled in the art will also know that, besides implementing the controller in the form of purely computer-readable program code, the same functions can be achieved by logically programming the method steps, making the controller take the form of logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers (PLCs), and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the devices included within it for implementing various functions can also be considered structures within that hardware component. Alternatively, the devices for implementing various functions can be considered as both software modules implementing the method and structures within a hardware component.

[0079] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0080] For ease of description, the above apparatus is described by dividing it into various functional units. Of course, when implementing one or more embodiments of this specification, the functions of each unit can be implemented in one or more software and / or hardware.

[0081] Those skilled in the art will understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects.

[0082] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0083] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0084] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1The steps of the function specified in one or more boxes.

[0085] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0086] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0087] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0088] This specification can be described in the general context of computer-executable instructions that are executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This specification can also be practiced in distributed computing environments, where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside on local and remote computer storage media, including storage devices.

[0089] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0090] The above description is merely an embodiment of this specification and is not intended to limit the scope of one or more embodiments of this specification. Various modifications and variations can be made to one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of one or more embodiments of this specification.

Claims

1. A method for generating washing strategies based on visual recognition, characterized in that, Specifically, the following steps are included: Acquire a first image set of clothes to be washed; wherein the first image set is an overall image of multiple stacked clothes acquired by an image acquisition device deployed in the garment care device; The first image set is input into the clothing management database pre-stored in the clothing care device for matching to obtain a first clothing information set; wherein, the first clothing information set includes at least washing interaction information corresponding to each individual garment to be washed; the washing interaction information includes at least the overall image of the individual garment to be washed and its corresponding first clothing material information and preset washing strategy information. Acquire a second image set of the garments to be washed; wherein the second image set consists of detailed images of the stained areas of each individual garment acquired by an image acquisition device deployed within the garment care device; The second image set is input into a pre-trained local washing inference model for inference to obtain a second clothing information set; The second image set and the second clothing information set are input into the first clothing information set for retrieval, so as to match the first clothing material information and preset washing strategy information corresponding to each piece of clothing to be washed in the second image set.

2. The washing strategy generation method based on visual recognition according to claim 1, characterized in that, The first clothing information set also includes the total number of the clothing items to be washed.

3. The washing strategy generation method based on visual recognition according to claim 1, characterized in that, The second clothing information set includes clothing parameter information, stain data information, and suggested washing strategy information corresponding to the individual garment.

4. The washing strategy generation method based on visual recognition according to claim 3, characterized in that, The clothing parameter information includes second clothing material information corresponding to the detailed image of the stained area of ​​the single garment; the stain data information includes at least one or more of the following: clothing stain type, clothing stain area, and clothing stain intensity.

5. The washing strategy generation method based on visual recognition according to claim 1, characterized in that, It also includes the following steps: Determine whether the suggested washing strategy information matches the preset washing strategy information; If the suggested washing strategy information matches the preset washing strategy information, then the suggested washing strategy information is executed; If the suggested washing strategy information does not match the preset washing strategy information, the preset washing strategy information is adjusted using the stain data information to generate the washing strategy information to be executed.

6. The washing strategy generation method based on visual recognition according to claim 1, characterized in that, It also includes the following steps: The second clothing information set is transmitted back to the user through interaction; Determine whether the second clothing information set needs to be modified; If correction is required, the user-corrected second clothing information set is input into the first clothing information set for retrieval; If no correction is required, the second clothing information set is input into the first clothing information set for retrieval.

7. The washing strategy generation method based on visual recognition according to claim 1, characterized in that, It also includes the following steps: Determine whether it is necessary to combine cloud servers for reasoning; If it is necessary to combine the cloud server for inference, the second image set is input into the remote washing inference model deployed in the cloud server for inference to obtain the second clothing information set; If inference is not required using the cloud server, the second image set is input into a pre-trained local washing inference model to obtain the second clothing information set.

8. The washing strategy generation method based on visual recognition according to claim 7, characterized in that, When inference is required in conjunction with the cloud server, the following steps are also included: The second set of clothing information is fed back to the local washing inference model to optimize the local washing inference model.

9. The washing strategy generation method based on visual recognition according to claim 1, characterized in that, The first image set is input into the clothing management database pre-stored in the clothing care device for matching to obtain the first clothing information set, including the following steps: Multiple clothing regions of the overall image of the stacked clothing are obtained as the target clothing image set for washing; The individual images from the set of images of the target clothes to be washed are sequentially input into the clothes management database; The clothing features of the single image are compared one by one with the images in the clothing management database; Determine if the match was successful; If the match is successful, retrieve the first clothing information set; If the matching fails, the single image is input into the pre-trained local washing inference model for inference to obtain the first clothing information set; The first set of clothing information and the single image are stored in the clothing management database.

10. The washing strategy generation method based on visual recognition according to claim 9, characterized in that, The clothing features correspond to the information set classification stored in the clothing management database, and include at least one or more of clothing color, clothing type, and clothing size.

11. The washing strategy generation method based on visual recognition according to claim 1, characterized in that, The second image set and the second clothing information set are input into the first clothing information set for retrieval, in order to match the first clothing material information and preset washing strategy information corresponding to each individual piece of clothing to be washed in the second image set, including the following steps: Obtain an overall image of a single garment from the first garment information set; Acquire a multi-target region image of the entire image of the single garment; wherein, the multi-target region image consists of multiple images containing stained areas; Obtain the multi-target region image and the local feature map in the second image set; Local feature maps in the second image set are retrieved one by one from the local feature images of the multi-target region image to obtain candidate target images; Retrieve the first clothing material information of the candidate target image; The second clothing material information is input into the first clothing material information for retrieval, so as to match the first clothing material information corresponding to the second clothing information; Obtain the preset washing strategy information corresponding to the first clothing material information.

12. A washing strategy generation system based on visual recognition, characterized in that, Includes the following modules: An image acquisition module is configured to acquire a first image set of clothes to be washed; wherein the first image set is an overall image of multiple stacked clothes acquired by an image acquisition device deployed within the garment care device; The image acquisition module is also used to acquire a second image set of the clothes to be washed; wherein, the second image set is a detailed image of the stained area of ​​each individual garment acquired by the image acquisition device deployed in the garment care device; The information acquisition module is configured to input the first image set into a clothing management database pre-stored in the clothing care device for matching, so as to obtain a first clothing information set; wherein, the first clothing information set includes at least washing interaction information corresponding to each individual garment to be washed; the washing interaction information includes at least the overall image corresponding to each individual garment to be washed and its corresponding first clothing material information and preset washing strategy information. The model inference module is configured to input the second image set into a pre-trained local washing inference model for inference to obtain a second clothing information set; The washing strategy generation module is configured to input the second image set and the second clothing information set into the first clothing information set for retrieval, so as to match the first clothing material information and preset washing strategy information corresponding to each piece of clothing to be washed in the second image set. The cloud server module is configured to deploy a pre-trained remote washing inference model.

13. The washing strategy generation system based on visual recognition according to claim 12, characterized in that, The image acquisition module includes a wide-angle camera unit and a macro camera unit; wherein, the wide-angle camera unit is used to acquire a first set of images of the clothes to be washed, and the macro camera unit is used to acquire a second set of images of the clothes to be washed.

14. A readable storage medium having a computer program stored thereon, characterized in that, The computer program is executed by a processor using the visual recognition-based washing strategy generation method as described in any one of claims 1-11.

15. An electronic device, characterized in that, include: Processor and memory, the memory being used to store one or more programs; When the processor executes the one or more programs, it implements the visual recognition-based washing strategy generation method as described in any one of claims 1-11.

Images