Resource recycling intelligent management method and device and electronic equipment

Abstract

The invention discloses a resource recycling intelligent management method and device and electronic equipment, and relates to the field of image processing. The method comprises the following steps: acquiring the actual weight of a to-be-detected commodity, and scanning the to-be-detected commodity to obtain a commodity number and a first commodity category; querying the standard weight of the to-be-detected commodity in a preset weight library; calculating a difference value between the actual weight and the standard weight; if the difference value is smaller than or equal to a preset difference threshold value, the to-be-detected commodity is conveyed to an intelligent classification area; photographing the to-be-detected commodity to obtain a commodity image; performing feature extraction on the commodity image to obtain commodity features, and classifying the to-be-detected commodities to obtain a second commodity category; judging whether the first commodity category is matched with the second commodity category; and if it is determined that the first commodity category is matched with the second commodity category, sorting the to-be-detected commodity to a sorting area corresponding to the first commodity category. According to the technical scheme, the weight serves as the primary screening index, the abnormal commodities are intercepted at the source, and the sorting accuracy is improved.

Description

Technical Field

[0001] The present application relates to the field of image recognition, and specifically to a resource recycling and intelligent management method, device and electronic equipment. Background Art

[0002] With the booming development of global e-commerce and the increasing diversification of consumer demand, sorting centers play an increasingly important role in supply chain management. Modern sorting centers not only need to handle huge quantities of goods from different regions, but also need to ensure that the goods are sorted and delivered to the correct destination quickly and accurately. In order to meet these needs, sorting centers have begun to adopt various intelligent technologies, such as automated weight measurement, scanning and recognition systems, and image processing technology to improve processing speed and accuracy.

[0003] At present, despite the introduction of automated and semi-automated systems, the relevant sorting centers still face some problems. For example, the relevant technology uses a single image processing method when sorting, but some goods may have abnormal weights due to internal defects without any abnormality on the outside. Therefore, the relevant technology has the problem of low recognition accuracy.

[0004] Therefore, there is an urgent need for an intelligent management method, device and electronic equipment for resource recycling. Summary of the invention

[0005] The present application provides a resource recycling intelligent management method, device and electronic equipment, which uses weight as a preliminary screening indicator to intercept abnormal goods at the source, reduce the occurrence of incorrect sorting, and improve recognition accuracy.

[0006] In a first aspect of the present application, a method for intelligent management of resource recycling is provided, the method comprising: obtaining the actual weight of a commodity to be inspected, scanning the commodity to be inspected to obtain a commodity number and a first commodity category; querying the standard weight of the commodity to be inspected in a preset weight library according to the commodity number; calculating the difference between the actual weight of the commodity to be inspected and the standard weight, and judging the relationship between the difference and a preset difference threshold; if the difference is less than or equal to the preset difference threshold, transmitting the commodity to be inspected to an intelligent classification area; in the intelligent classification area, photographing the commodity to be inspected to obtain a commodity image; extracting features from the commodity image to obtain commodity features, and classifying the commodity to be inspected according to the commodity features to obtain a second commodity category; judging whether the first commodity category and the second commodity category match; if it is determined that the first commodity category and the second commodity category match, sorting the commodity to be inspected to the sorting area corresponding to the first commodity category.

[0007] By adopting the above technical solution, by obtaining the actual weight and standard weight of the goods to be inspected, calculating the difference between the two and judging the size relationship between the difference and the preset difference threshold, the goods with abnormal weight can be quickly screened out, reducing the workload of subsequent intelligent classification. For goods with normal weight, by taking pictures to obtain the product image, and extracting and classifying the product image, the actual category of the goods to be inspected can be automatically determined, and matched and verified with the expected category. The matching goods can directly enter the corresponding sorting area to realize automatic and intelligent sorting processing. This technical solution makes full use of the two attributes of weight and image to establish a multi-dimensional product abnormal screening and category verification mechanism. Weight, as a preliminary screening indicator, can intercept a batch of abnormal goods at the source to prevent them from entering the complex image classification process, thereby improving overall efficiency. Through this method, the sorting center can greatly improve the level of automation and processing efficiency, reduce manual intervention and misclassification and omission, and improve the accuracy of sorting.

[0008] Optionally, after querying the standard weight of the product to be inspected in the preset weight library according to the product number, the method also includes: if there is no corresponding standard weight information in the preset weight library, weighing multiple target products to obtain the target weight corresponding to each of the target products, the product category of the target product is the first product category; adding the multiple target weights and taking the average value to obtain the average weight; using the average weight as the standard weight, and storing the correspondence between the average weight and the first product category in the preset weight library.

[0009] By adopting the above technical solution, when it is found that a certain product number has no corresponding record in the standard weight library, the learning process can be automatically triggered. By weighing multiple target products of the same category and taking the average, the standard weight of the product to be inspected is generated and stored in the preset weight library. In this way, the standard weight library can be continuously expanded and improved to adapt to the ever-changing types of products, with stronger adaptability and practicality.

[0010] Optionally, photographing the commodity to be inspected to obtain a commodity image specifically includes: photographing the commodity to be inspected by using multiple cameras at different angles to obtain multi-perspective sub-commodity images; and splicing the sub-commodity images to obtain a commodity image of the commodity to be inspected.

[0011] By adopting the above technical solution, a multi-angle camera is used to take pictures of the goods to be inspected, and multiple sub-product images are stitched together to obtain a more comprehensive and clearer product image. Compared with shooting from a single angle, multi-angle can cover different sides and details of the product, providing richer visual information. Through image stitching, these local features can be integrated into a complete product image, which is convenient for subsequent feature extraction and category judgment. Multi-angle shooting can deal with problems such as the unfixed placement angle and partial occlusion of the goods to be inspected, ensuring that the key features of the goods to be inspected can be captured from at least several angles.

[0012] Optionally, classifying the goods to be inspected according to the product characteristics to obtain a second product category specifically includes: inputting the product characteristics into an input layer of a preset product training model, and calculating probability values ​​corresponding to each product category through a forward propagation algorithm; sorting each of the probability values, and taking the product category corresponding to the maximum probability value among the multiple probability values ​​as the second product category.

[0013] By adopting the above technical solution and using the preset commodity classification model to classify and judge the extracted commodity features, the commodity category of the commodity to be inspected can be automatically and efficiently identified. Using this method, the workload of manual verification can be greatly reduced, and the sorting efficiency and accuracy can be improved.

[0014] Optionally, before classifying the goods to be inspected according to the goods characteristics to obtain the second goods category, the method also includes: acquiring historical image data of different goods categories, constructing an image data set, and dividing the image data set to obtain training set data, verification set data, and test set data; using the training set data to train the preset goods training model, and using the verification set data to evaluate the training effect to obtain the accuracy rate; judging whether the accuracy rate is greater than or equal to a preset accuracy rate threshold; if it is determined that the accuracy rate is greater than or equal to the preset accuracy rate threshold, stopping the training.

[0015] By adopting the above technical solution, a comprehensive and balanced image sample space can be constructed by collecting historical image data of different commodity categories and dividing them into training sets, validation sets and test sets according to a certain ratio. The classification model is trained with the training set data so that it can learn to extract commodity features and classification rules from a large number of samples. The model performance is evaluated with the validation set data, and the accuracy is used as the optimization target, which can guide the iterative optimization of the model and improve its generalization ability and classification accuracy.

[0016] Optionally, after calculating the difference between the actual weight of the goods to be inspected and the standard weight, and determining the size relationship between the difference and a preset difference threshold, the method further includes: if the difference is greater than the preset difference threshold, marking the goods to be inspected as goods with abnormal weight, and transferring the goods to be inspected to a manual review area.

[0017] By adopting the above technical solution, for goods whose actual weight differs greatly from the standard value, they can be automatically marked as abnormal and transferred to the manual review area as a priority. This can timely detect goods with abnormal weight and prevent them from entering the subsequent automatic sorting process, thereby controlling misclassification and error sorting at the source.

[0018] Optionally, after determining whether the first commodity category and the second commodity category match, the method further includes: if it is determined that the first commodity category and the second commodity category do not match, transferring the commodity to be inspected to a manual review area.

[0019] By adopting the above technical solution, for those products with normal weight but the second product category obtained by image classification does not match the first product category, they can be automatically transferred to the manual review area for review by specialized quality inspectors, thereby effectively identifying and isolating products with image classification errors or uncertainties, and preventing misclassification and omissions. Through manual review, the image classification results can be verified and corrected to improve the accuracy of product identification.

[0020] In a second aspect of the present application, a resource recycling intelligent management device is provided, the device comprising: an acquisition module and a processing module, wherein: the acquisition module is used to acquire the actual weight of the commodity to be inspected, and scan the commodity to be inspected to obtain the commodity number and the first commodity category; the processing module is used to query the standard weight of the commodity to be inspected in a preset weight library according to the commodity number; the processing module is also used to calculate the difference between the actual weight of the commodity to be inspected and the standard weight, and judge the relationship between the difference and the preset difference threshold; the processing module is also used to check the actual weight of the commodity to be inspected if the difference is less than or equal to the preset difference threshold. value threshold, the commodity to be inspected is transmitted to the intelligent classification area; the processing module is further used to take a photo of the commodity to be inspected in the intelligent classification area to obtain a commodity image; the processing module is further used to extract features from the commodity image to obtain commodity features, and classify the commodity to be inspected according to the commodity features to obtain a second commodity category; the processing module is further used to determine whether the first commodity category and the second commodity category match; the processing module is further used to sort the commodity to be inspected to the sorting area corresponding to the first commodity category if it is determined that the first commodity category and the second commodity category match.

[0021] In the third aspect of the present application, an electronic device is provided, including a processor, a memory, a user interface and a network interface, the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device executes any one of the methods described above.

[0022] In a fourth aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores instructions, and when the instructions are executed, any of the methods described above is executed.

[0023] In summary, one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. By obtaining the actual weight and standard weight of the goods to be inspected, calculating the difference between the two and judging the size relationship between the difference and the preset difference threshold, the goods with abnormal weight can be quickly screened out, reducing the workload of subsequent intelligent classification. For goods with normal weight, the product image is obtained by taking a photo, and the feature extraction and classification of the product image can be performed to automatically determine the actual category of the goods to be inspected, and match and verify it with the expected category. The matching goods can directly enter the corresponding sorting area to realize automatic and intelligent sorting processing. This technical solution makes full use of the two attributes of weight and image to establish a multi-dimensional product abnormal screening and category verification mechanism. Weight, as a preliminary screening indicator, can intercept a batch of abnormal goods at the source to prevent them from entering the complex image classification process, thereby improving overall efficiency. Through this method, the sorting center can greatly improve the level of automation and processing efficiency, reduce manual intervention and misclassification and omission, and improve the accuracy of sorting.

[0024] 2. Use a multi-angle camera to take pictures of the goods to be inspected, and stitch multiple sub-product images together to obtain a more comprehensive and clearer product image. Compared with shooting from a single angle, multi-angle can cover different sides and details of the product, providing richer visual information. Through image stitching, these local features can be integrated into a complete product image, which is convenient for subsequent feature extraction and category judgment. Multi-angle shooting can deal with problems such as the unfixed placement angle and partial occlusion of the goods to be inspected, ensuring that the key features of the goods to be inspected can be captured from at least several angles.

[0025] 3. By collecting historical image data of different commodity categories and dividing them into training sets, validation sets, and test sets in a certain proportion, a comprehensive and balanced image sample space can be constructed. The classification model is trained using the training set data so that it can learn to extract commodity features and classification rules from a large number of samples. The model performance is evaluated using the validation set data, and accuracy is used as the optimization target, which can guide the iterative optimization of the model and improve its generalization ability and classification accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a flow chart of a resource recycling intelligent management method disclosed in an embodiment of the present application; Figure 2 It is a module schematic diagram of a resource recycling intelligent management device disclosed in an embodiment of the present application; Figure 3 It is a structural schematic diagram of an electronic device disclosed in an embodiment of the present application.

[0027] Explanation of the reference numerals: 201, acquisition module; 202, processing module; 300, electronic device; 301, processor; 302, communication bus; 303, user interface; 304, network interface; 305, memory. DETAILED DESCRIPTION

[0028] In order to enable technicians in this field to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments.

[0029] In the description of the embodiments of the present application, words such as "for example" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "for example" or "for example" is intended to present related concepts in a specific way.

[0030] In the description of the embodiments of the present application, the meaning of the term "multiple" refers to two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. The terms "include", "comprise", "have" and their variations all mean "including but not limited to", unless otherwise specifically emphasized.

[0031] This application provides a resource recycling intelligent management method, referring to Figure 1 , Figure 11 is a flow chart of a resource recycling intelligent management method provided in an embodiment of the present application. The method is applied to a server, and the server is used to execute a sorting program. The server can be a single server, or a server cluster composed of multiple servers, or a cloud computing service center. The method includes steps S101 to S108, and the steps are as follows: Step S101: obtaining the actual weight of the commodity to be inspected, and scanning the commodity to be inspected to obtain the commodity number and the first commodity category.

[0032] In step S101, when the goods to be inspected arrive at the unloading platform of the sorting center, the server controls the automatic guided vehicle (AGV) to transport the goods to the intelligent weighing workbench. The intelligent weighing workbench is equipped with a high-precision electronic scale and a barcode scanner, which can automatically complete the collection of the weight of the goods and the recognition of the barcode.

[0033] The server establishes a communication connection with the intelligent weighing workbench and receives the data information sent by the intelligent weighing workbench in real time. When the goods to be inspected are placed on the electronic scale, the server controls the electronic scale to weigh the goods to be inspected and uploads the collected actual weight to the server. At the same time, the server controls the barcode scanner to scan the barcode on the outer packaging of the goods to obtain the barcode information.

[0034] After receiving the actual weight and barcode information, the server will parse and process the data. For the actual weight, the server will associate it with the product number to form a structured weight record and store it in the database. For the barcode information, the server will parse it according to the preset coding rules and extract the product number and the first product category.

[0035] Step S102: According to the product number, the standard weight of the product to be inspected is searched in the preset weight database.

[0036] In step S102, the server pre-establishes and maintains a preset weight library, which stores the standard weight information of various commodities. The preset weight library can be stored in a relational database or a non-relational database, and each record in the preset weight library contains fields such as product number, product name, product category, standard weight, etc. The initial data of the preset weight library comes from the standard weight data provided by the commodity supplier or is obtained through statistical analysis of a large amount of actual weighing data of historical commodities. When the server obtains the actual weight and product number of the commodity to be inspected from the intelligent weighing workbench, it will immediately initiate a query request in the preset weight library. The server uses the product number as a keyword to search for the standard weight that matches it in the preset weight library.

[0037] In a possible implementation, after step S102, the method further includes: if there is no corresponding standard weight information in the preset weight library, weighing multiple target commodities to obtain a target weight corresponding to each target commodity, the commodity category of the target commodity being the first commodity category; adding the multiple target weights and taking the average value to obtain an average weight; using the average weight as the standard weight, and storing the correspondence between the average weight and the first commodity category in the preset weight library.

[0038] Specifically, when there is no standard weight record corresponding to the number of the product to be inspected in the preset weight library, a standard weight generation process will be automatically triggered. The server first searches the database for historical product records of the same category as the first product category of the product to be inspected. These historical products are the target products, which belong to the same product category as the product to be inspected. After the server finds the target products, it obtains the actual weight one by one, and the weighing data comes from the historical weight records in the database.

[0039] For example, the server did not find the standard weight of beverage A with the number "69012345" in the preset weight library, but found 5 other beverage products of the same category (such as beverage B, beverage C, beverage D, beverage E, and beverage F) by querying historical data. The server obtained the actual weights of these 5 target products, which were 502 grams, 498 grams, 505 grams, 501 grams, and 499 grams respectively. After the server obtains the actual weight data of a series of target products, it will perform arithmetic averaging on these data to obtain an average weight value. In the above beverage product example, the sum of the actual weights of the 5 target products is 2505 grams, and the average weight is 2505 grams / 5=501 grams. The server uses this average weight value as the standard weight of the product to be inspected (beverage A).

[0040] Next, the server associates the generated standard weight with the first commodity category to form a new weight record, and stores it in the preset weight library.

[0041] Step S103: Calculate the difference between the actual weight of the commodity to be inspected and the standard weight, and determine the relationship between the difference and a preset difference threshold.

[0042] In step S103, the server obtains the actual weight of the commodity to be inspected from step S101, and obtains the standard weight corresponding to the commodity to be inspected from step S102. The server performs a difference operation on the two weight values ​​to obtain the difference between the actual weight and the standard weight. After calculating the difference, the server compares it with the preset difference threshold to determine whether the weight of the commodity to be inspected is within a reasonable range. The preset difference threshold can be set according to factors such as commodity category, weight level, packaging material, etc., and this application does not limit this.

[0043] After step S03, the method further includes: if the difference is greater than a preset difference threshold, marking the commodity to be inspected as a commodity with abnormal weight, and transferring the commodity to be inspected to a manual review area.

[0044] Specifically, when the server compares the difference obtained in step S102 with the preset difference threshold, if the difference is greater than the preset difference threshold, it is determined that the weight of the commodity to be inspected is abnormal, and the server marks the commodity as abnormal weight. For example, the server finds the information record of the commodity to be inspected in the database and updates the "weight status" field from "normal" to "abnormal". After the marking is completed, the server will control the sorting system to transfer the commodity to be inspected to the manual review area. The manual review area is a special area set up inside the sorting center for manual review and processing of abnormal commodities screened out by the sorting system.

[0045] The server will first send a control command to the sorting system, instructing the product to be inspected to be removed from the normal sorting process and directed to the manual review area. For example, when a product with abnormal weight passes a key node on the sorting line, the server will control the turnaround gate of the node in real time to direct the product with abnormal weight to a dedicated conveyor line and finally send it to the manual review area.

[0046] After arriving at the manual review area, abnormal products will be temporarily stored in a special buffer area. At the same time, the server will send a notification to the staff in the manual review area, prompting that there are new abnormal products that need to be processed. Notification can be achieved in many ways, such as listing the abnormal products on the display screen in the manual review area, or pushing messages to the handheld terminal devices of the staff.

[0047] Step S104: If the difference is less than or equal to the preset difference threshold, the commodity to be inspected is transmitted to the intelligent classification area.

[0048] In step S104, when the server compares the difference obtained in step S102 with the preset difference threshold, if the difference is less than or equal to the preset difference threshold, it is determined that the weight of the commodity to be inspected is normal, and the server marks the commodity to be inspected as normal weight. Similar to the commodity with abnormal weight, the server implements this marking process by modifying the status field in the commodity information record. After marking, the server will control the sorting system to continue to transmit the commodity to be inspected with normal weight in the normal sorting process, and finally guide it to the intelligent classification area.

[0049] Step S105: In the intelligent classification area, take photos of the goods to be inspected to obtain product images.

[0050] In step S105, the commodity to be inspected is photographed by multiple cameras at different angles to obtain sub-commodity images of multiple viewing angles; the sub-commodity images are spliced ​​to obtain a commodity image of the commodity to be inspected.

[0051] Specifically, when the goods to be inspected enter the intelligent classification area, the server triggers the photo-taking process. Multiple high-resolution digital cameras are installed in the intelligent classification area, and they are deployed at different angles and positions to achieve all-round shooting of the goods. The server establishes a connection with the digital camera through the network to control the shooting parameters and triggering timing in real time. When the goods to be inspected pass through the shooting area of ​​each digital camera, the server triggers the corresponding digital camera to shoot in real time. Each digital camera captures a partial image of the product, that is, the sub-product image, according to its own perspective and position. The sub-product image shows the various faces and details of the goods to be inspected from different angles. After the server collects a group of sub-product images from each digital camera, it will transmit them to the image processing module 202 for splicing. Image splicing is a computer vision technology. In the embodiment of the present application, it can be understood as using the overlapping areas and feature points between the sub-product images of the goods to be inspected to seamlessly stitch multiple partial images into a complete product image.

[0052] In the process of stitching product images, the server first preprocesses the sub-product images, such as image correction and denoising. The server uses feature point detection and matching algorithms, such as SIFT, to find common feature points between adjacent sub-product images and calculate the mapping relationship between them. Based on the feature point mapping relationship, the server calculates the stitching transformation matrix between the sub-product images, and then uses the stitching transformation matrix to spatially transform and reproject the sub-product images so that they are aligned in a common coordinate system. Finally, the server fuses the aligned sub-product images to generate the final product image.

[0053] Step S106: extracting features from the commodity image to obtain commodity features, and classifying the commodity to be inspected according to the commodity features to obtain a second commodity category.

[0054] Before step S106, the method also includes: acquiring historical image data of different commodity categories, constructing an image data set, and dividing the image data set to obtain training set data, verification set data, and test set data; using the training set data to train a preset commodity training model, and using the verification set data to evaluate the training effect to obtain the accuracy rate; determining whether the accuracy rate is greater than or equal to a preset accuracy rate threshold; if it is determined that the accuracy rate is greater than or equal to the preset accuracy rate threshold, stopping the training.

[0055] Specifically, the server obtains a large amount of historical commodity image data for constructing an image dataset. The historical image data comes from actual sorting operation records or specially collected sample images. The server retrieves and extracts historical image data of different commodity categories from the image storage database; when obtaining historical image data, the server will perform necessary preprocessing and cleaning on the historical image data, such as filtering out samples with poor image quality and unrecognizable, adjusting the image size and resolution, and performing image enhancement and standardization. The server organizes and annotates the acquired and preprocessed historical image data according to commodity categories to construct a structured image dataset. Each image data is associated with the commodity category label of the commodity.

[0056] After building the image dataset, the server divides it into training set, validation set, and test set according to a certain ratio. The division ratio is 70-80% for the training set, 10-15% for the validation set, and 10% for the test set. For example, the server randomly selects 80% of the samples from the image dataset as the training set, and then randomly selects 10% from the remaining samples as the validation set, and finally the remaining 10% as the test set. After the dataset is prepared, the server trains the preset product recognition model. The preset product recognition model uses a convolutional neural network (CNN), which can automatically extract multi-level features in the image and judge the product category based on these features.

[0057] The server inputs the training set data into the preset product recognition model, which calculates the prediction results through forward propagation, and then adjusts the model parameters through back propagation to continuously reduce the error between the prediction results and the actual labels. This process is continuously iterated, allowing the preset product recognition model to learn product features and classification rules from the training data.

[0058] The server will determine whether the accuracy of the preset product recognition model on the validation set has reached the preset accuracy threshold. The preset accuracy threshold is set based on actual business needs and model performance, such as 90%, 95%, etc. If the current accuracy is greater than or equal to the preset accuracy threshold, the server can stop training and consider that the model has reached the expected performance level.

[0059] If the current accuracy does not reach the preset accuracy threshold, the server will continue to train the preset accuracy threshold until the accuracy meets the requirement. After the preset product recognition model is trained, the server will use the test set data to evaluate the performance of the preset product recognition model. The test set is a set of completely independent samples that the model has not been exposed to during the training process. The accuracy on the test set can objectively measure the actual performance and generalization ability of the model.

[0060] Through the above process, the server completes the construction of the image data set and the training and optimization of the preset product recognition model. The trained preset product recognition model can effectively identify the category of newly taken product images and provide a reliable judgment basis for intelligent sorting.

[0061] In step S106, the product features are input into the input layer of the preset product training model, and the probabilities corresponding to each product category are calculated through the forward propagation algorithm; the probability values ​​are sorted, and the product category corresponding to the maximum probability value among the multiple probability values ​​is taken as the second product category.

[0062] Specifically, the server inputs the product image generated by splicing in step S105 into a preset product recognition model. The preset product recognition model consists of multiple convolutional layers, pooling layers and fully connected layers, which can automatically extract multi-level features in the product image and classify the products to be inspected according to these features.

[0063] The input layer of the preset product recognition model receives the pixel data of the product image, and transmits and transforms the image data layer by layer through the forward propagation algorithm. In the convolution layer, the preset product recognition model uses multiple convolution kernels to perform convolution operations on the product image to extract local features. In the pooling layer, the preset product recognition model downsamples the convolution features to reduce the size of the feature map while retaining the most significant feature information. The local features gradually form higher-level and advanced features through the stacking of multiple convolution layers and pooling layers.

[0064] In the last few fully connected layers of the preset product recognition model, high-level features are further integrated and transformed to form a fixed-length feature vector, namely the product feature. The product feature is a compressed representation of the entire product image, which contains the key feature information of the product. After obtaining the product feature, the server inputs it to the output layer of the preset product recognition model. The output layer maps the product feature to the probability distribution of each predefined product category. The output value of the output layer represents the probability that the product to be inspected belongs to this product category.

[0065] The server sorts the probability values ​​given by the output layer and finds the maximum probability value. The commodity category corresponding to the maximum probability value is used as the second commodity category determined by the preset commodity recognition model.

[0066] Step S107: Determine whether the first commodity category and the second commodity category match.

[0067] In step S107, the server compares the first commodity category and the second commodity category to determine whether they are the same, ie, match.

[0068] After step S107, the method further includes: if it is determined that the first commodity category and the second commodity category do not match, transferring the commodity to be inspected to a manual review area.

[0069] Specifically, if the server determines that the first commodity category and the second commodity category do not match, it means that the actual attributes of the commodity do not match the expectations, and there may be problems such as input errors and packaging errors, and manual review is required. The server will control the sorting system to transfer the commodity to be inspected to the manual review area.

[0070] Step S108: If it is determined that the first commodity category and the second commodity category match, the commodities to be inspected are sorted to a sorting area corresponding to the first commodity category.

[0071] Specifically, if the server determines that the first commodity category matches the second commodity category, it means that the actual attributes of the commodity are consistent with expectations and can be processed according to the normal sorting process. The server will control the sorting system to sort the commodity to be inspected to the sorting area corresponding to the first commodity category. The sorting area is a physical area in the warehouse used to store commodities of the same category or the same destination. Each commodity category has its own corresponding sorting area.

[0072] Reference Figure 2 The present application also provides a resource recycling intelligent management device, which is a server, and the server includes an acquisition module 201 and a processing module 202, wherein: the acquisition module 201 is used to obtain the actual weight of the commodity to be inspected, and scan the commodity to be inspected to obtain the commodity number and the first commodity category; the processing module 202 is used to query the standard weight of the commodity to be inspected in the preset weight library according to the commodity number; the processing module 202 is also used to calculate the difference between the actual weight and the standard weight of the commodity to be inspected, and judge the size relationship between the difference and the preset difference threshold; the processing module 202 is also used if the difference If the difference between the first commodity category and the second commodity category is less than or equal to the preset difference threshold, the commodity to be inspected is transmitted to the intelligent classification area; the processing module 202 is also used to take a photo of the commodity to be inspected in the intelligent classification area to obtain a commodity image; the processing module 202 is also used to extract features from the commodity image to obtain commodity features, and classify the commodity to be inspected according to the commodity features to obtain a second commodity category; the processing module 202 is also used to determine whether the first commodity category and the second commodity category match; the processing module 202 is also used to sort the commodity to be inspected to the sorting area corresponding to the first commodity category if it is determined that the first commodity category and the second commodity category match.

[0073] In a possible implementation, after the processing module 202 queries the standard weight of the product to be inspected in the preset weight library according to the product number, the method also includes: if there is no corresponding standard weight information in the preset weight library, the processing module 202 weighs multiple target products to obtain the target weight corresponding to each target product, and the product category of the target product is the first product category; the processing module 202 adds the multiple target weights and takes the average value to obtain the average weight; the processing module 202 uses the average weight as the standard weight, and stores the correspondence between the average weight and the first product category in the preset weight library.

[0074] In a possible implementation, the processing module 202 takes a photo of the product to be inspected to obtain a product image, which specifically includes: the processing module 202 takes a photo of the product to be inspected using multiple cameras at different angles to obtain multi-perspective sub-product images; the processing module 202 splices the sub-product images to obtain a product image of the product to be inspected.

[0075] In a possible implementation, the processing module 202 classifies the inspected goods according to the goods features to obtain a second product category, specifically including: the processing module 202 inputs the product features into the input layer of a preset product training model, and calculates the probability values ​​corresponding to each product category through a forward propagation algorithm; the processing module 202 sorts the probability values, and takes the product category corresponding to the maximum probability value among the multiple probability values ​​as the second product category.

[0076] In a possible implementation, the processing module 202 classifies the inspected goods according to the goods characteristics, and before obtaining the second goods category, the method also includes: the acquisition module 201 obtains historical image data of different goods categories, constructs an image data set, and divides the image data set to obtain training set data, verification set data, and test set data; the processing module 202 uses the training set data to train a preset goods training model, and uses the verification set data to evaluate the training effect to obtain the accuracy rate; the processing module 202 determines whether the accuracy rate is greater than or equal to a preset accuracy rate threshold; if it is determined that the accuracy rate is greater than or equal to the preset accuracy rate threshold, the processing module 202 stops training.

[0077] In one possible implementation, after the processing module 202 calculates the difference between the actual weight of the goods to be inspected and the standard weight, and determines the relationship between the difference and a preset difference threshold, the method further includes: if the difference is greater than the preset difference threshold, the processing module 202 marks the goods to be inspected as goods with abnormal weight, and transfers the goods to be inspected to a manual review area.

[0078] In a possible implementation, after the processing module 202 determines whether the first commodity category and the second commodity category match, the method further includes: if the processing module 202 determines that the first commodity category and the second commodity category do not match, transferring the commodity to be inspected to a manual review area.

[0079] It should be noted that: when the device provided in the above embodiment realizes its function, only the division of the above functional modules is used as an example. In actual application, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0080] The present application also provides an electronic device. Figure 3 , Figure 3 The electronic device 300 may include: at least one processor 301 , at least one network interface 304 , a user interface 303 , a memory 305 , and at least one communication bus 302 .

[0081] The communication bus 302 is used to realize the connection and communication between these components.

[0082] The user interface 303 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 303 may also include a standard wired interface and a wireless interface.

[0083] The network interface 304 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0084] Among them, the processor 301 may include one or more processing cores. The processor 301 uses various interfaces and lines to connect various parts in the entire server, and executes various functions of the server and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 305, and calling data stored in the memory 305. Optionally, the processor 301 can be implemented in at least one hardware form of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), and programmable logic array (Programmable Logic Array, PLA). The processor 301 can integrate one or a combination of a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU) and a modem. Among them, the CPU mainly processes the operating system, user interface and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 301, and it can be implemented separately through a chip.

[0085] Among them, the memory 305 may include a random access memory (Random Access Memory, RAM) and may also include a read-only memory (Read-Only Memory). Optionally, the memory 305 includes a non-transitory computer-readable storage medium. The memory 305 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 305 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 305 may optionally also be at least one storage device located away from the aforementioned processor 301. Refer to Figure 3 , the memory 305 as a computer storage medium may include an operating system, a network communication module, a user interface module and an application program of a resource recycling and intelligent management method.

[0086] exist Figure 3In the electronic device 300 shown, the user interface 303 is mainly used to provide an input interface for the user and obtain the data input by the user; and the processor 301 can be used to call the application program storing a resource recycling and utilization intelligent management method in the memory 305, and when executed by one or more processors 301, the electronic device 300 executes one or more of the methods described in the above embodiments. It should be noted that for the aforementioned method embodiments, for the sake of simple description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required for the present application.

[0087] The present application also provides a computer-readable storage medium, which stores instructions. When executed by one or more processors 301, the electronic device 300 executes one or more of the methods described in the above embodiments.

[0088] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0089] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0090] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0091] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0092] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a memory and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned memory includes: various media that can store program codes, such as USB flash drives, mobile hard drives, magnetic disks or optical disks.

[0093] The above is only an exemplary embodiment of the present disclosure and cannot be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the disclosure of the specification and the truth of practice, those skilled in the art will easily think of other embodiments of the present disclosure.

[0094] This application is intended to cover any variation, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the art not described in the present disclosure. The description and examples are to be regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A resource recycling intelligent management method, characterized in that: The method comprises: Obtaining the actual weight of the commodity to be inspected, and scanning the commodity to be inspected to obtain the commodity number and the first commodity category; According to the commodity number, query the standard weight of the commodity to be inspected in the preset weight library; Calculating the difference between the actual weight of the commodity to be inspected and the standard weight, and determining the relationship between the difference and a preset difference threshold; If the difference is less than or equal to the preset difference threshold, the commodity to be inspected is transmitted to the intelligent classification area; In the intelligent classification area, taking photos of the commodities to be inspected to obtain commodity images; Extracting features from the commodity image to obtain commodity features, and classifying the commodity to be inspected according to the commodity features to obtain a second commodity category; Determining whether the first commodity category and the second commodity category match; If it is determined that the first commodity category and the second commodity category match, the commodity to be inspected is sorted to a sorting area corresponding to the first commodity category.

2. The method according to claim 1, characterized in that After searching the preset weight library for the standard weight of the commodity to be inspected according to the commodity number, the method further includes: If there is no corresponding standard weight information in the preset weight library, a plurality of target commodities are weighed to obtain a target weight corresponding to each of the target commodities, and the commodity category of the target commodities is the first commodity category; Adding a plurality of the target weights and taking an average value to obtain an average weight; The average weight is used as the standard weight, and the corresponding relationship between the average weight and the first commodity category is stored in the preset weight library.

3. The method according to claim 1, characterized in that The taking a photo of the commodity to be inspected to obtain a commodity image specifically includes: The commodity to be inspected is photographed by multiple cameras at different angles to obtain sub-commodity images from multiple perspectives; The sub-product images are spliced ​​to obtain a product image of the product to be inspected.

4. The method according to claim 1, characterized in that: The classifying the commodity to be inspected according to the commodity characteristics to obtain a second commodity category specifically includes: Input the product features into the input layer of the preset product training model, and calculate the probability value corresponding to each product category through the forward propagation algorithm; The probability values ​​are sorted, and the commodity category corresponding to the maximum probability value among the plurality of probability values ​​is taken as the second commodity category.

5. The method according to claim 4, characterized in that Before classifying the commodity to be inspected according to the commodity characteristics to obtain the second commodity category, the method further includes: Acquire historical image data of different commodity categories, construct an image dataset, and divide the image dataset into training set data, validation set data, and test set data; The training set data is used to train the preset product training model, and the validation set data is used to evaluate the training effect to obtain the accuracy rate; Determining whether the accuracy is greater than or equal to a preset accuracy threshold; If it is determined that the accuracy is greater than or equal to the preset accuracy threshold, the training is stopped.

6. The method according to claim 1, characterized in that After calculating the difference between the actual weight of the commodity to be inspected and the standard weight, and determining the magnitude relationship between the difference and a preset difference threshold, the method further includes: If the difference is greater than the preset difference threshold, the commodity to be inspected is marked as a commodity with abnormal weight, and the commodity to be inspected is transferred to a manual review area.

7. The method according to claim 1, characterized in that After determining whether the first commodity category and the second commodity category match, the method further includes: If it is determined that the first commodity category and the second commodity category do not match, the commodity to be inspected is transferred to a manual review area.

8. A resource recycling intelligent management device, characterized in that: The device comprises an acquisition module (201) and a processing module (202), wherein: The acquisition module (201) is used to acquire the actual weight of the commodity to be inspected, and scan the commodity to be inspected to obtain the commodity number and the first commodity category; The processing module (202) is used to query the standard weight of the commodity to be inspected in a preset weight library according to the commodity number; The processing module (202) is further used to calculate the difference between the actual weight of the commodity to be inspected and the standard weight, and determine the relationship between the difference and a preset difference threshold; The processing module (202) is further configured to transmit the commodity to be inspected to an intelligent classification area if the difference is less than or equal to the preset difference threshold; The processing module (202) is further used to take a photo of the commodity to be inspected in the intelligent classification area to obtain a commodity image; The processing module (202) is further used to extract features from the commodity image to obtain commodity features, and classify the commodity to be inspected according to the commodity features to obtain a second commodity category; The processing module (202) is further used to determine whether the first commodity category and the second commodity category match; The processing module (202) is further configured to sort the commodity to be inspected into a sorting area corresponding to the first commodity category if it is determined that the first commodity category matches the second commodity category.

9. An electronic device, characterized in that: The electronic device (300) comprises a processor (301), a memory (305), a user interface (303) and a network interface (304), wherein the memory (305) is used to store instructions, the user interface (303) and the network interface (304) are used to communicate with other devices, and the processor (301) is used to execute the instructions stored in the memory (305) so that the electronic device (300) executes the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed, the method according to any one of claims 1 to 7 is executed.

Images