Electric energy meter component identification method

Abstract

The invention relates to an electric energy meter component identification method and device, computer equipment, a computer readable storage medium and a computer program product, and can be applied to the technical field of artificial intelligence. The method comprises the following steps: dividing an electric energy meter component sample image; creating a model training item; training a to-be-trained electric energy meter component identification model by using the training parameters and the target training set; performing verification processing on the trained electric energy meter component identification model by using the target verification set, and performing test processing on the trained electric energy meter component identification model by using the test set; taking the trained electric energy meter component identification model as a target electric energy meter component identification model; and performing electric energy meter component identification on the to-be-identified electric energy meter component image by using the target electric energy meter component identification model to obtain an electric energy meter component identification result of the to-be-identified electric energy meter component image. By adopting the method, the identification efficiency of the electric energy meter component can be improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and particularly to a method, device, computer device, computer-readable storage medium, and computer program product for identifying components of an electric energy meter. Background Art

[0002] With the rapid development of artificial intelligence technology, computer vision has been increasingly widely applied in various fields. Among them, object detection and image recognition technologies, as important branches of computer vision, play an increasingly important role in industrial inspection, quality control, etc. Therefore, how to efficiently identify components of an electric energy meter has become an important research direction.

[0003] Traditional technologies usually identify components of an electric energy meter by means of manual visual inspection; however, identifying components of an electric energy meter in this way requires a lot of manual processing time, resulting in low efficiency of identifying components of an electric energy meter. Summary of the Invention

[0004] Based on this, in view of the above technical problems, it is necessary to provide a method, device, computer device, computer-readable storage medium, and computer program product for identifying components of an electric energy meter that can improve the efficiency of identifying components of an electric energy meter.

[0005] In a first aspect, the present application provides a method for identifying components of an electric energy meter. The method includes:

[0006] Obtain a sample image of components of an electric energy meter;

[0007] Divide the sample image of components of an electric energy meter into a training and validation set to be labeled and a test set, and perform labeling processing on the training and validation set to be labeled to obtain a target training and validation set; the target training and validation set includes a target training set and a target validation set;

[0008] Upload the target training and validation set and the test set to a model training platform;

[0009] Create a model training project corresponding to the target training and validation set and the test set on the model training platform;

[0010] Update the training parameters corresponding to the model training project, and use the updated training parameters and the target training set to train an electric energy meter component identification model to be trained to obtain a trained electric energy meter component identification model;

[0011] Use the target verification set to verify the trained electricity meter component recognition model to obtain the verification result of the trained electricity meter component recognition model, and use the test set to test the trained electricity meter component recognition model to obtain the test result of the trained electricity meter component recognition model;

[0012] When both the verification result and the test result indicate passing, use the trained electricity meter component recognition model as the target electricity meter component recognition model;

[0013] Use the target electricity meter component recognition model to identify the electricity meter components in the electricity meter component image to be recognized, and obtain the electricity meter component recognition result of the electricity meter component image to be recognized.

[0014] In one embodiment, the using the target electricity meter component recognition model to identify the electricity meter components in the electricity meter component image to be recognized and obtaining the electricity meter component recognition result of the electricity meter component image to be recognized includes:

[0015] Obtain the electricity meter component image to be recognized;

[0016] Input the electricity meter component image to be recognized into the target electricity meter component recognition model for recognition, and obtain the category information and position information of the electricity meter components in the electricity meter component image to be recognized;

[0017] Determine the electricity meter component recognition result of the electricity meter component image to be recognized according to the category information and position information of the electricity meter components.

[0018] In one embodiment, after using the target electricity meter component recognition model to identify the electricity meter components in the electricity meter component image to be recognized and obtaining the electricity meter component recognition result of the electricity meter component image to be recognized, it further includes:

[0019] Determine the category information and position information of the faulty components in the electricity meter component image according to the electricity meter component recognition result of the electricity meter component image to be recognized;

[0020] Determine the fault location result of the electricity meter component image to be recognized according to the category information and position information of the faulty components.

[0021] In one embodiment, the using the target verification set to verify the trained electricity meter component recognition model to obtain the verification result of the trained electricity meter component recognition model includes:

[0022] Input the target validation set into the trained electric energy meter component recognition model for recognition to obtain the recognition result of the target validation set;

[0023] Determine the verification index information corresponding to the recognition result of the target validation set according to the target validation set;

[0024] Determine the verification result according to the verification index information.

[0025] In one embodiment, the obtaining of the electric energy meter component sample image includes:

[0026] Control an image acquisition device to perform multi-dimensional shooting on a plurality of circuit board samples to obtain a plurality of circuit board sample images;

[0027] Extract the image area containing the electric energy meter components from the plurality of circuit board sample images to obtain the electric energy meter component sample image.

[0028] In one embodiment, after taking the trained electric energy meter component recognition model as the target electric energy meter component recognition model when both the verification result and the test result indicate passing, it further includes:

[0029] Perform export and packaging processing on the target electric energy meter component recognition model to obtain the target electric energy meter component recognition model to be deployed;

[0030] Deploy the target electric energy meter component recognition model to be deployed to a target server.

[0031] In a second aspect, the present application also provides an electric energy meter component recognition device. The device includes:

[0032] An image acquisition module, configured to acquire an electric energy meter component sample image;

[0033] An image division module, configured to divide the electric energy meter component sample image into a training and validation set and a test set to be labeled, and perform labeling processing on the training and validation set to be labeled to obtain a target training and validation set; the target training and validation set includes a target training set and a target validation set;

[0034] A target upload module, configured to upload the target training and validation set and the test set to a model training platform;

[0035] A project creation module, configured to create a model training project corresponding to the target training and validation set and the test set on the model training platform;

[0036] A model training module, configured to update training parameters corresponding to the model training project, and use the updated training parameters and the target training set to train the electricity meter component recognition model to be trained, so as to obtain a trained electricity meter component recognition model;

[0037] A model verification module, configured to perform verification processing on the trained electricity meter component recognition model by using the target verification set, obtain a verification result of the trained electricity meter component recognition model, and perform test processing on the trained electricity meter component recognition model by using the test set, so as to obtain a test result of the trained electricity meter component recognition model;

[0038] A model determination module, configured to use the trained electricity meter component recognition model as the target electricity meter component recognition model when both the verification result and the test result indicate passing;

[0039] An image recognition module, configured to perform electricity meter component recognition on an electricity meter component image to be recognized by using the target electricity meter component recognition model, so as to obtain an electricity meter component recognition result of the electricity meter component image to be recognized.

[0040] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, the following steps are implemented:

[0041] Obtain an electricity meter component sample image;

[0042] Divide the electricity meter component sample image into a training and verification set to be labeled and a test set, and perform labeling processing on the training and verification set to be labeled to obtain a target training and verification set; the target training and verification set includes a target training set and a target verification set;

[0043] Upload the target training and verification set and the test set to a model training platform;

[0044] Create a model training project corresponding to the target training and verification set and the test set on the model training platform;

[0045] Update training parameters corresponding to the model training project, and use the updated training parameters and the target training set to train the electricity meter component recognition model to be trained, so as to obtain a trained electricity meter component recognition model;

[0046] Use the target validation set to perform a validation process on the trained electricity meter component recognition model to obtain the validation result of the trained electricity meter component recognition model, and use the test set to perform a test process on the trained electricity meter component recognition model to obtain the test result of the trained electricity meter component recognition model;

[0047] When both the validation result and the test result indicate passing, use the trained electricity meter component recognition model as the target electricity meter component recognition model;

[0048] Use the target electricity meter component recognition model to perform electricity meter component recognition on the electricity meter component image to be recognized, and obtain the electricity meter component recognition result of the electricity meter component image to be recognized.

[0049] Fourthly, the present application also provides a computer-readable storage medium. The computer-readable storage medium has a computer program stored thereon, and when the computer program is executed by a processor, the following steps are implemented:

[0050] Obtain an electricity meter component sample image;

[0051] Divide the electricity meter component sample image into a training and validation set to be labeled and a test set, and perform a labeling process on the training and validation set to be labeled to obtain a target training and validation set; the target training and validation set includes a target training set and a target validation set;

[0052] Upload the target training and validation set and the test set to a model training platform;

[0053] Create a model training project corresponding to the target training and validation set and the test set on the model training platform;

[0054] Update the training parameters corresponding to the model training project, and use the updated training parameters and the target training set to train a to-be-trained electricity meter component recognition model to obtain a trained electricity meter component recognition model;

[0055] Use the target validation set to perform a validation process on the trained electricity meter component recognition model to obtain the validation result of the trained electricity meter component recognition model, and use the test set to perform a test process on the trained electricity meter component recognition model to obtain the test result of the trained electricity meter component recognition model;

[0056] When both the validation result and the test result indicate passing, use the trained electricity meter component recognition model as the target electricity meter component recognition model;

[0057] Use the target electric energy meter component recognition model to perform electric energy meter component recognition on the image of the electric energy meter component to be recognized, and obtain the recognition result of the electric energy meter component of the image of the electric energy meter component to be recognized.

[0058] In a fifth aspect, the present application also provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0059] Obtain a sample image of an electric energy meter component;

[0060] Divide the sample image of the electric energy meter component into a training and validation set to be labeled and a test set, and perform labeling processing on the training and validation set to be labeled to obtain a target training and validation set; the target training and validation set includes a target training set and a target validation set;

[0061] Upload the target training and validation set and the test set to a model training platform;

[0062] Create a model training project corresponding to the target training and validation set and the test set on the model training platform;

[0063] Update the training parameters corresponding to the model training project, and use the updated training parameters and the target training set to train the electric energy meter component recognition model to be trained, and obtain a trained electric energy meter component recognition model;

[0064] Use the target validation set to perform validation processing on the trained electric energy meter component recognition model to obtain the validation result of the trained electric energy meter component recognition model, and use the test set to perform test processing on the trained electric energy meter component recognition model to obtain the test result of the trained electric energy meter component recognition model;

[0065] When both the validation result and the test result indicate passing, use the trained electric energy meter component recognition model as the target electric energy meter component recognition model;

[0066] Use the target electric energy meter component recognition model to perform electric energy meter component recognition on the image of the electric energy meter component to be recognized, and obtain the recognition result of the electric energy meter component of the image of the electric energy meter component to be recognized.

[0067] The above method, device, computer equipment, computer-readable storage medium and computer program product for identifying components of an electric energy meter acquire a sample image of the components of the electric energy meter; divide the sample image of the components of the electric energy meter into a training and validation set to be labeled and a test set, and perform a labeling process on the training and validation set to be labeled to obtain a target training and validation set; the target training and validation set includes a target training set and a target validation set; upload the target training and validation set and the test set to a model training platform; create a model training project corresponding to the target training and validation set and the test set on the model training platform; update the training parameters corresponding to the model training project, and use the updated training parameters and the target training set to train the electric energy meter component identification model to be trained to obtain a trained electric energy meter component identification model; use the target validation set to perform a validation process on the trained electric energy meter component identification model to obtain a validation result of the trained electric energy meter component identification model, and use the test set to perform a test process on the trained electric energy meter component identification model to obtain a test result of the trained electric energy meter component identification model; in the case where both the validation result and the test result indicate passing, use the trained electric energy meter component identification model as the target electric energy meter component identification model; use the target electric energy meter component identification model to perform electric energy meter component identification on the image of the electric energy meter component to be identified to obtain an electric energy meter component identification result of the image of the electric energy meter component to be identified. This solution is conducive to constructing high-quality training data by acquiring a sample image of the components of the electric energy meter and performing data set division and labeling; is conducive to improving the learning effect of the model by creating a project on the model training platform, optimizing the training parameters, and using the target training set for model training; is conducive to comprehensively evaluating the model performance and ensuring the generalization ability of the model by using the validation set and the test set for validation and testing respectively; finally, using the evaluated model for actual identification tasks is conducive to improving the efficiency and accuracy of identifying components of the electric energy meter. Description of the Drawings

[0068] To more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following will briefly introduce the drawings required for use in the description of the embodiments or related technologies. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0069] Figure 1 It is a schematic flowchart of a method for identifying components of an electric energy meter in an embodiment;

[0070] Figure 2 It is a schematic flowchart of the steps for determining the identification result of the components of the electric energy meter in an embodiment;

[0071] Figure 3 is a schematic flow chart of a method for identifying components of an electric energy meter in another embodiment;

[0072] Figure 4 is a structural block diagram of a device for identifying components of an electric energy meter in one embodiment;

[0073] Figure 5 is an internal structure diagram of a computer device in one embodiment. Detailed implementation manners

[0074] In order to make the objectives, technical solutions and advantages of the present application clearer and more understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0075] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data need to comply with relevant regulations.

[0076] In an exemplary embodiment, as Figure 1 shown, a method for identifying components of an electric energy meter is provided. In this embodiment, the method is exemplified by being applied to a terminal; it can be understood that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. Among them, the terminal can be but is not limited to various personal computers, laptop computers, smart phones, tablet computers, etc.; the server can be an independent physical server, can also be a server cluster or distributed system composed of multiple physical servers, and can also be a cloud server providing cloud computing services. In this embodiment, the method includes the following steps:

[0077] Step S101, obtaining a sample image of an electric energy meter component.

[0078] Step S102, dividing the sample image of the electric energy meter component into a training and validation set to be labeled and a test set, and performing labeling processing on the training and validation set to be labeled to obtain a target training and validation set; the target training and validation set includes a target training set and a target validation set.

[0079] Step S103, uploading the target training and validation set and the test set to a model training platform.

[0080] Step S104, creating a model training project corresponding to the target training and validation set and the test set on the model training platform.

[0081] Step S105: Update the training parameters corresponding to the model training project, and use the updated training parameters and the target training set to train the electricity meter component recognition model to be trained, obtaining the trained electricity meter component recognition model.

[0082] Step S106: Use the target validation set to perform a validation process on the trained electricity meter component recognition model, obtaining the validation result of the trained electricity meter component recognition model, and use the test set to perform a test process on the trained electricity meter component recognition model, obtaining the test result of the trained electricity meter component recognition model.

[0083] Step S107: In the case where both the validation result and the test result indicate passing, use the trained electricity meter component recognition model as the target electricity meter component recognition model.

[0084] Step S108: Use the target electricity meter component recognition model to perform electricity meter component recognition on the electricity meter component image to be recognized, obtaining the electricity meter component recognition result of the electricity meter component image to be recognized.

[0085] Among them, the electricity meter component sample image can be image data containing electricity meter components. For example, the electricity meter component sample image can be a large number of images containing components collected from different circuit board samples. These images are taken from all directions, at multiple angles, and at multiple distances to ensure coverage of all component types to be recognized.

[0086] Among them, the training and validation set can be a data set used for model training and validation. For example, the training and validation set can be a data set composed of randomly selected 90% of the pictures from the electricity meter component sample images.

[0087] Among them, the test set can be an independent data set used to evaluate the model performance. For example, the test set can be a data set composed of randomly selected 10% of the pictures from the electricity meter component sample images.

[0088] Among them, the annotation process can be a process of marking and classifying the target objects in the image. For example, the annotation process can be using annotation tools to annotate the pictures in the training and validation set, annotating the components to be detected, including thermistors, liquid crystal drivers, transformers, Bluetooth modules, three-terminal voltage regulators, power chips, metering chips, CPU chips, chip capacitors, optocouplers, liquid crystal displays, etc.

[0089] Among them, the target training and validation set can be the training and validation set after the annotation process. For example, the target training and validation set can be a data set divided into a target training set and a target validation set according to a ratio of 9:1.

[0090] Among them, the model training platform can be a software platform for deep learning model training. For example, the model training platform can be the PaddlePaddle (Ai Studio) platform.

[0091] Among them, the model training project can be a working environment created on the model training platform for training a specific model. For example, the model training project can be a project created on the PaddlePaddle (Ai Studio) platform.

[0092] Among them, the training parameters can be various settings for configuring and controlling the model training process. For example, the training parameters can include the learning rate, hyperparameters of the optimization function, etc.

[0093] Among them, the electric energy meter component recognition model can be a deep learning model for recognizing electric energy meter components.

[0094] Among them, the verification process can be a process of evaluating the model performance using a validation set. For example, the verification process can be calculating metrics such as precision, recall, accuracy, and mean average precision of the model on the target validation set.

[0095] Among them, the testing process can be a process of evaluating the model generalization ability using an independent test set. For example, the testing process can be using the test set to perform inference on the trained model, checking the inference results, and preventing model overfitting.

[0096] Among them, the target electric energy meter component recognition model can be the final model that has been trained, verified, and tested and whose performance meets the requirements.

[0097] Among them, the image of the electric energy meter component to be recognized can be a new electric energy meter image that needs to be component-recognized. For example, the image of the electric energy meter component to be recognized can be an image of an electric energy meter circuit board taken in actual applications.

[0098] Among them, the recognition result of the electric energy meter component can be the detection and classification output of the model for the components in the image to be recognized. For example, the recognition result of the electric energy meter component can include information such as the recognized component type, position coordinates, and confidence level.

[0099] Optionally, the terminal collects a large number of images containing components from different circuit board samples, taking pictures from all directions, multiple angles, and multiple distances to ensure that all types of components to be identified are covered, and obtains the sample images of the components of the electricity meter. The terminal randomly divides the obtained sample images of the components of the electricity meter into a training and validation set to be labeled and a test set according to the ratio of 90% and 10%. The terminal uses a professional labeling tool to label the training and validation set to be labeled, labeling the components to be detected, including thermistors, liquid crystal drivers, transformers, Bluetooth modules, three-terminal voltage regulators, power chips, metering chips, CPU chips, chip capacitors, optocouplers, liquid crystal displays, etc., to obtain the target training and validation set. The terminal divides the target training and validation set into a target training set and a target validation set according to the ratio of 9:1. The terminal uploads the target training and validation set and the test set to the PaddlePaddle platform and creates a model training project on this platform. The terminal modifies the configuration file corresponding to the algorithm, updates the training parameters such as the learning rate and the hyperparameters of the optimization function, and uses these parameters and the target training set to train the electricity meter component recognition model to be trained. The terminal uses the target validation set to calculate the precision, recall rate, accuracy, mean average precision and other indicators of the trained electricity meter component recognition model to obtain the verification result. The terminal uses the test set to perform inference on the trained electricity meter component recognition model, views the inference result, and obtains the test result. When both the verification result and the test result meet the preset standards, the terminal exports the trained electricity meter component recognition model as an inference model, which is used as the target electricity meter component recognition model. The terminal uses the target electricity meter component recognition model to identify the components in the image of the electricity meter circuit board taken in actual application, and outputs information such as the identified component type, position coordinates, and confidence as the electricity meter component recognition result.

[0100] In the above method for identifying components of an electric energy meter, a sample image of the components of the electric energy meter is obtained; the sample image of the components of the electric energy meter is divided into a training and validation set to be labeled and a test set, and the training and validation set to be labeled is labeled to obtain a target training and validation set; the target training and validation set includes a target training set and a target validation set; the target training and validation set and the test set are uploaded to a model training platform; a model training project corresponding to the target training and validation set and the test set is created on the model training platform; the training parameters corresponding to the model training project are updated, and the to-be-trained component identification model of the electric energy meter is trained using the updated training parameters and the target training set to obtain a trained component identification model of the electric energy meter; the trained component identification model of the electric energy meter is verified using the target validation set to obtain a verification result of the trained component identification model of the electric energy meter, and the trained component identification model of the electric energy meter is tested using the test set to obtain a test result of the trained component identification model of the electric energy meter; in the case where both the verification result and the test result indicate passing, the trained component identification model of the electric energy meter is used as the target component identification model of the electric energy meter; the target component identification model of the electric energy meter is used to identify the components of the electric energy meter in the to-be-identified image of the components of the electric energy meter to obtain the identification result of the components of the electric energy meter in the to-be-identified image of the components of the electric energy meter. This solution is conducive to constructing high-quality training data by obtaining a sample image of the components of the electric energy meter and performing data set division and labeling; it is conducive to improving the learning effect of the model by creating a project on the model training platform, optimizing the training parameters, and using the target training set for model training; it is conducive to comprehensively evaluating the model performance and ensuring the generalization ability of the model by using the validation set and the test set for verification and testing respectively; finally, using the evaluated model for actual identification tasks is conducive to improving the efficiency and accuracy of identifying the components of the electric energy meter.

[0101] In an exemplary embodiment, with reference to Figure 2 , using the target component identification model of the electric energy meter to identify the components of the electric energy meter in the to-be-identified image of the components of the electric energy meter to obtain the identification result of the components of the electric energy meter in the to-be-identified image of the components of the electric energy meter specifically includes the following content:

[0102] Step S201, obtain the to-be-identified image of the components of the electric energy meter;

[0103] Step S202, input the to-be-identified image of the components of the electric energy meter into the target component identification model of the electric energy meter for identification to obtain the category information and position information of the components of the electric energy meter in the to-be-identified image of the components of the electric energy meter;

[0104] Step S203, determine the identification result of the components of the electric energy meter in the to-be-identified image of the components of the electric energy meter according to the category information and position information of the components of the electric energy meter.

[0105] Among them, the category information can be the type or kind of the identified electricity meter components. For example, the category information can include thermistors, liquid crystal drivers, transformers, Bluetooth modules, three-terminal voltage regulators, power chips, metering chips, CPU chips, chip capacitors, optocouplers, or liquid crystal displays, etc.

[0106] Among them, the position information can be the spatial position of the identified electricity meter components in the image. For example, the position information can include parameters such as the center coordinates (x, y), width (w), and height (h) of the components.

[0107] Optionally, the terminal acquires the image of the electricity meter component to be identified and inputs the image of the electricity meter component to be identified into the target electricity meter component recognition model for recognition. The terminal processes the image of the electricity meter component to be identified through the target electricity meter component recognition model, and obtains the category information and position information of the electricity meter components in the image of the electricity meter component to be identified. The terminal determines the recognition result of the electricity meter components in the image of the electricity meter component to be identified according to the category information and position information of the electricity meter components.

[0108] For example, the terminal captures the image of the electricity meter circuit board to be identified through a camera device or reads it from a database as the image of the electricity meter component to be identified. The terminal inputs the image of the electricity meter component to be identified into the target electricity meter component recognition model exported through the development kit. The target electricity meter component recognition model processes the input image, identifies each electricity meter component in the image, and outputs the category information (such as thermistors, liquid crystal drivers, transformers, etc.) and position information (such as center coordinates x and y, width w, height h) of each identified component. The terminal generates a list containing the type, position coordinates, and confidence level of each identified component as the recognition result of the electricity meter components in the image of the electricity meter component to be identified according to the category information and position information output by the model.

[0109] The technical solution provided in this embodiment realizes the automatic recognition and positioning of electricity meter components by inputting the image of the electricity meter component to be identified into the trained target electricity meter component recognition model, which is beneficial to quickly obtain the category and position information of the components, and thus is beneficial to improving the efficiency and accuracy of electricity meter component recognition.

[0110] In an exemplary embodiment, after using the target electric energy meter component recognition model to recognize the electric energy meter component image to be recognized and obtaining the recognition result of the electric energy meter component in the image to be recognized, the following steps are further included: determining the category information and location information of the faulty component in the electric energy meter component image according to the recognition result of the electric energy meter component in the image to be recognized; and determining the fault location result of the electric energy meter component image to be recognized according to the category information and location information of the faulty component.

[0111] Among them, the faulty component can be a component that appears abnormal or fails in the electric energy meter. For example, the faulty component can be a burned resistor, a damaged chip, or a short-circuited capacitor, etc.

[0112] Among them, the fault location result can be the determination of the specific location and type of the faulty component in the electric energy meter. For example, the fault location result can include information such as the type of the faulty component, the precise coordinates on the circuit board, and the cause of the fault.

[0113] Optionally, the terminal analyzes the characteristics of each recognized component according to the recognition result of the electric energy meter component in the image to be recognized. The terminal determines the category information and location information of the faulty component in the electric energy meter component image by comparing with a preset normal component parameter library. The terminal determines the fault location result of the electric energy meter component image to be recognized according to the category information and location information of the faulty component, in combination with the circuit schematic diagram and component layout diagram of the electric energy meter.

[0114] The technical solution provided in this embodiment determines the category and location information of the faulty component by analyzing the recognition result of the electric energy meter component, so as to perform fault location, which is beneficial to quickly and accurately identify and locate the faulty component in the electric energy meter, and thus is beneficial to improving the efficiency and accuracy of electric energy meter fault recognition.

[0115] In an exemplary embodiment, using a target validation set to perform a validation process on the trained electric energy meter component recognition model to obtain the validation result of the trained electric energy meter component recognition model specifically includes the following steps: inputting the target validation set into the trained electric energy meter component recognition model for recognition to obtain the recognition result of the target validation set; determining the validation index information corresponding to the recognition result of the target validation set according to the target validation set; and determining the validation result according to the validation index information.

[0116] Among them, the validation index information can be a quantitative index used to evaluate the model performance. For example, the validation index information can include precision, recall rate, average precision, etc.

[0117] Optionally, the terminal inputs the target validation set into the trained electricity meter component recognition model for recognition to obtain the recognition results of the target validation set. The terminal calculates validation metric information, including precision, recall, and mean average precision, based on the true annotation information and recognition results of the target validation set. The terminal compares the calculated validation metric information with a preset performance threshold to determine the validation result of the trained electricity meter component recognition model.

[0118] The technical solution provided in this embodiment is beneficial to comprehensively evaluate the model performance and timely discover problems existing in the model by using the target validation set to validate the trained model and calculating the validation metric information, thereby facilitating the optimization of the model and improving the accuracy of electricity meter component recognition.

[0119] In an exemplary embodiment, obtaining the electricity meter component sample images specifically includes the following: controlling an image acquisition device to perform multi-dimensional shooting on a plurality of circuit board samples to obtain a plurality of circuit board sample images; extracting image regions containing electricity meter components from the plurality of circuit board sample images to obtain the electricity meter component sample images.

[0120] Among them, the image acquisition device can be a hardware device for acquiring circuit board images.

[0121] Among them, the circuit board samples can be circuit boards for training and testing the electricity meter component recognition model. For example, the circuit board samples can be electricity meter circuit boards of different models and different production batches.

[0122] Among them, multi-dimensional shooting can be to shoot the circuit board samples from different angles, different distances, and different lighting conditions. For example, multi-dimensional shooting can include shooting from multiple angles such as the front, side, and top views, close and far distances, and different lighting conditions such as natural light and artificial light.

[0123] Among them, the circuit board sample images can be the original images obtained through multi-dimensional shooting. For example, the circuit board sample images can be high-resolution pictures containing the entire circuit board.

[0124] Optionally, the terminal controls the image acquisition device to perform multi-dimensional shooting on a plurality of circuit board samples to obtain circuit board sample images under different angles, distances, and lighting conditions. The terminal uses image processing algorithms to preprocess the obtained circuit board sample images, including image denoising, contrast adjustment, and color correction. The terminal extracts image regions containing electricity meter components from the preprocessed circuit board sample images to generate the electricity meter component sample images.

[0125] The technical solution provided in this embodiment is beneficial to obtaining the sample images of the electricity meter components under multiple angles and multiple lighting conditions by performing multi-dimensional shooting and image region extraction on the circuit board samples, which is conducive to increasing the diversity and representativeness of the samples, and thus is conducive to improving the generalization ability and recognition accuracy of the electricity meter component recognition model.

[0126] In an exemplary embodiment, after the trained electricity meter component recognition model is used as the target electricity meter component recognition model when both the verification result and the test result indicate passing, the following steps are further included: exporting and packaging the target electricity meter component recognition model to obtain the target electricity meter component recognition model to be deployed; deploying the target electricity meter component recognition model to be deployed to the target server.

[0127] Among them, the exporting and packaging process can be a process of converting the trained model into a deployable format. For example, the exporting and packaging process can include steps such as serialization, optimization, and compression of the model structure and weights.

[0128] Among them, the target electricity meter component recognition model to be deployed can be a model file after the exporting and packaging process. For example, the target electricity meter component recognition model to be deployed can be a model file containing model structure and weight information.

[0129] Among them, the target server can be a computer system for running the electricity meter component recognition service.

[0130] Optionally, after the terminal confirms that both the verification result and the test result indicate passing, the trained electricity meter component recognition model is determined as the target electricity meter component recognition model. The terminal uses a model export tool to perform an export operation on the target electricity meter component recognition model, converting the model structure and weights into a deployable format. The terminal performs optimization and compression processing on the exported model to generate the target electricity meter component recognition model to be deployed. The terminal transmits the target electricity meter component recognition model to be deployed to the target server through a network connection and completes the installation and configuration of the model on the target server.

[0131] The technical solution provided in this embodiment is beneficial to converting the trained model into a service that can run in the actual environment by exporting, packaging, and deploying the model that has been verified and tested, realizing the engineering application of the model, and thus is beneficial to improving the automation degree and practicality of electricity meter component recognition.

[0132] The following uses an application example to illustrate the electricity meter component recognition method provided in this application. This application example takes the application of this method to the terminal as an example.

[0133] As electric meter products become smaller and more complex, the design of printed circuit boards (PCBs) becomes more and more intensive. At the same time, with the development of the power grid, the number of electric meters put into operation is also increasing. The large number of electric meters also brings a huge amount of electric meter fault inspection and maintenance work.

[0134] Traditional manual identification methods require high professional skills of personnel. Identifying electrical components by visually inspecting circuit boards requires the inspectors to be familiar with the model parameters of electrical components, etc. Faced with a large number of complex types of electric meters, a large number of staff are required to participate in the inspection.

[0135] Traditional computer image recognition methods rely on feature extraction and matching. Feature extraction methods include SIFT (Scale Invariant Feature Transform), SURF (Speeded Up Robust Features), ORB (Oriented Fast and Rotation Invariant Features), etc. These algorithms can extract key points and feature vectors from images and realize component recognition through feature matching. It extracts local feature points and their descriptors in the image, uses the descriptors for feature matching, and determines the similarity of images.

[0136] Template matching is also a basic image recognition method, which compares the image to be recognized with a predefined template to find the area with the highest matching degree. Matching algorithms include correlation-based matching (such as cross-correlation, normalized cross-correlation) and distance-based matching (such as Euclidean distance). It uses a sliding window to search for areas similar to the template in the image, and calculates the similarity between the template and the local area of ​​the image to find the matching position.

[0137] Statistical methods use the global statistical characteristics of images for recognition, such as principal component analysis (PCA) and linear discriminant analysis (LDA). These methods reduce the dimension and project the image data to find the optimal separation between different categories. They calculate the statistical characteristics or transformation coefficients of the image and use these characteristics for classification and recognition.

[0138] The existing technologies have the following deficiencies to a greater or lesser extent:

[0139] Traditional manual visual recognition is slow and error-prone. It cannot meet the requirements of rapid recognition of a large number of electric meters. When a new model of electric energy meter appears, it is necessary to train the inspection personnel, which takes a long time and is costly.

[0140] The feature extraction and matching algorithm development cycle is long, and it is necessary to summarize and extract the features of each component and relationship features in the meter circuit board, which is a lot of work. In actual use, the feature extraction and matching algorithm has poor robustness to lighting, occlusion and complex backgrounds, and has high requirements for the consistency of the captured images, which easily leads to the problem of low recognition rate of electronic components. At the same time, the extraction and matching of feature points requires a large amount of calculation, which requires high computing resources.

[0141] Template matching requires pre - defining matching templates. The design and selection of templates need manual intervention, which requires a high level of personnel technical skills and a large amount of domain knowledge and experience. At the same time, different matching templates need to be designed for the circuit boards of different electric meters, resulting in a large workload. Since the position of the captured images may not be exactly the same in the development stage and the production environment, conditions such as the scale, rotation, and illumination of the images may cause abnormal situations such as a decrease in recognition rate and positioning errors, failing to meet the requirements.

[0142] Statistical analysis algorithms such as principal component analysis and linear discriminant analysis mainly find the optimal separation between different categories and require a large amount of training sample size. When some images have high noise, it is easy to reduce the recognition accuracy of the algorithms. When the shooting angle or lens parameters change, the slight deformation of the captured images will also lead to a decrease in the algorithm accuracy, and the algorithm has poor generalization adaptability.

[0143] This application example is based on the PaddlePaddle platform. Through the object detection of deep learning, it automatically identifies the components of the electric meter and locates faults, replacing the traditional manual identification and inspection of electric meter components to improve the accuracy and efficiency of the identification of electric meter components. Figure 3 , obtain the images of electric meter components, divide the images into a training - validation set and a test set, label various components of the electric meter in the training - validation set, upload the data set to the PaddlePaddle platform, create a project to prepare the training environment, adjust the parameter configuration and start training, model evaluation and model optimization, and model deployment and service provision.

[0144] Q1 (Step 1). Obtain the images of electric meter components: Collect a large number of images containing components from different circuit board samples, and take pictures from all directions, multiple angles, and multiple distances to ensure that all types of components to be recognized are covered.

[0145] Q2 (Step 2). Divide the images into a training - validation set and a test set and construct the coordinate system: For the images of electric meter components obtained in Q1, randomly select 10% of the images of each type as the test set for testing the quality of the model in the later stage; at the same time, divide the input graph into multiple grids. Each grid detects the objects falling into its area, each grid predicts a fixed number of bounding boxes and gives a confidence level. This box describes the rectangular area that may contain the object, including parameters such as the center x, center y, width w, and height h. And a class probability distribution is given along with the bounding box, indicating the probability that the target belongs to each class.

[0146] Q3 (Step 3). Annotate various components of the electricity meter in the training and validation set: Use the remaining images after Q2 extraction as the training and validation set, and use professional annotation tools (such as the first annotation tool LabelMe or the second annotation tool LabelImg) to annotate the components to be detected, including thermistors, liquid crystal drivers, transformers, Bluetooth modules, three-terminal regulators, power chips, metering chips, CPU chips, chip capacitors, optocouplers, liquid crystal displays, etc.

[0147] Q4 (Step 4). Upload the dataset to the PaddlePaddle platform: Convert the annotated dataset into the COCO (Common Objects in Context) dataset format, and divide it into a training set and a validation set according to a ratio of 9:1. Then, package it in separate directories together with the test set images and upload it to the dataset platform of the PaddlePaddle (Ai Studio) platform; then, the confidence score of the bounding box and the class prediction score are combined into the final score according to the algorithm, representing the accuracy of the box position and the credibility of the prediction.

[0148] Q5 (Step 5). Create a project and prepare the training environment: Enter the PaddlePaddle (Ai Studio) platform, select to create a project, fill in the project name, add the uploaded dataset, then select the JupyterLab (Jupiter Lab) IDE (Integrated Development Environment) and the project framework PaddlePaddle 2.6.1 (project framework PaddlePaddle 2.6.1), and click OK to complete the creation. Then select the created project, click to start the environment, select the advanced GPU V100 16G (Graphics Processing Unit V100 version with 16GB video memory) resource, click the OK button, wait for the page to prompt that the environment has been successfully started, and then click the Enter button to enter the PaddlePaddle training platform. Then unzip the uploaded dataset and install the PaddleDetection (PaddlePaddle Detection) development kit and the dependencies required for training.

[0149] Q6 (Step 6). Adjust the parameter configuration and start training: Enter the configuration directory under the PaddleDetection directory, modify the corresponding configuration files of the algorithm yolov3_darknet53_original_270e_coco.yml (yolov3 darknet 53 original 270 epochs COCO configuration file), including coco_detection.yml and yolov3_reader.yml (COCO detection configuration file and yolov3 reader configuration file). Then, specify the configuration file through the train.py file in the tools directory (training Python file in the tools directory) for training. Since image gridification and multi-bounding box prediction will overlap and cross-point to one object, the model uses the non-maximum suppression algorithm to overflow the bounding boxes with a higher overlap degree and a lower confidence level and retain the best results.

[0150] Q7 (Step 7). Model Optimization and Model Evaluation: If the training process shows poor results, immediately stop the model training, adjust the hyperparameters of the learning rate and optimization function in optimizer_270e.yml (the optimizer configuration file for 270 epochs), and then retrain the model; evaluate the trained model using the validation set, and calculate metrics such as precision, recall, accuracy, and mean average precision. At the same time, perform inference on the test set based on the trained model, check the inference results, and prevent model overfitting.

[0151] Q8 (Step 8). Model Deployment and Service Provision: Use export_model.py (the Python file for exporting the model in the tools directory of the PaddleDetection development kit) to export the model, then package and download the inference model, and finally deploy the model on the server. The deployed model can process circuit board images in real time, identify components, and output the identification results and confidence levels.

[0152] The technical solutions provided by this application example can at least achieve the following beneficial effects: Improved recognition speed: The recognition speed of electricity meter components can be increased to the second level, which is a huge improvement compared to traditional recognition methods, and can meet the need for quickly recognizing a large number of electricity meters, improving the efficiency of fault repair. Improved recognition accuracy: Through various data augmentation and sample balancing data preprocessing methods, the generalization ability of the model can be improved, and the recognition accuracy of electricity meter components is significantly improved compared to traditional recognition methods. Precise positioning of faulty components. Fully automated process: The model inference process does not require manual intervention, and the inference results are automatically pushed to the next process system to achieve pipeline operation and improve efficiency. Support for rapid iteration: For new models of electricity meters, transfer learning can be performed based on this model, with a relatively fast training speed and compatibility with historical electricity meter models. Compared with the training cycle of manual recognition, there is a huge improvement. Compared with traditional recognition algorithms, the development time is significantly reduced.

[0153] It should be understood that although the steps in the flowcharts involved in the above embodiments are shown in sequence according to the arrows, these steps do not necessarily have to be executed in the order indicated by the arrows. Unless clearly stated in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least some of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages. These steps or stages do not necessarily have to be executed at the same time, but can be executed at different times. The execution order of these steps or stages does not necessarily have to be sequential, but can be executed alternately or in turn with at least some of the steps or stages in other steps or other steps.

[0154] Based on the same inventive concept, an embodiment of this application further provides a device for identifying electrical energy meter components for implementing the above-mentioned method for identifying electrical energy meter components. The solution provided by this device for solving problems is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the electrical energy meter component identification device provided below can refer to the limitations on the electrical energy meter component identification method in the above text, and will not be elaborated here.

[0155] In an exemplary embodiment, as Figure 4 shown, a device for identifying electrical energy meter components is provided. The device 400 for identifying electrical energy meter components may include:

[0156] An image acquisition module 401, configured to acquire an image of an electrical energy meter component sample;

[0157] An image division module 402, configured to divide the image of the electrical energy meter component sample into a training and validation set to be labeled and a test set, and perform labeling processing on the training and validation set to be labeled to obtain a target training and validation set; the target training and validation set includes a target training set and a target validation set;

[0158] A target upload module 403, configured to upload the target training and validation set and the test set to a model training platform;

[0159] A project creation module 404, configured to create a model training project corresponding to the target training and validation set and the test set on the model training platform;

[0160] A model training module 405, configured to update the training parameters corresponding to the model training project, and use the updated training parameters and the target training set to train the electrical energy meter component identification model to be trained to obtain a trained electrical energy meter component identification model;

[0161] A model verification module 406, configured to use the target validation set to perform verification processing on the trained electrical energy meter component identification model to obtain a verification result of the trained electrical energy meter component identification model, and use the test set to perform test processing on the trained electrical energy meter component identification model to obtain a test result of the trained electrical energy meter component identification model;

[0162] A model determination module 407, configured to use the trained electrical energy meter component identification model as the target electrical energy meter component identification model when both the verification result and the test result indicate passing;

[0163] An image recognition module 408, configured to use the target electrical energy meter component identification model to perform electrical energy meter component identification on the image of the electrical energy meter component to be identified to obtain an electrical energy meter component identification result of the image of the electrical energy meter component to be identified.

[0164] In an exemplary embodiment, the image recognition module 408 is further configured to obtain an image of a component of an electricity meter to be recognized; input the image of the component of the electricity meter to be recognized into a target component recognition model of the electricity meter for recognition, and obtain the category information and position information of the component of the electricity meter in the image of the component of the electricity meter to be recognized; and determine the recognition result of the component of the electricity meter in the image of the component of the electricity meter to be recognized according to the category information and position information of the component of the electricity meter.

[0165] In an exemplary embodiment, the apparatus 400 further includes: a result determination module, configured to determine the category information and position information of a faulty component in the image of the component of the electricity meter according to the recognition result of the component of the electricity meter in the image of the component of the electricity meter to be recognized; and determine the fault location result of the image of the component of the electricity meter to be recognized according to the category information and position information of the faulty component.

[0166] In an exemplary embodiment, the model verification module 406 is further configured to input a target verification set into the trained component recognition model of the electricity meter for recognition, and obtain the recognition result of the target verification set; determine the verification index information corresponding to the recognition result of the target verification set according to the target verification set; and determine the verification result according to the verification index information.

[0167] In an exemplary embodiment, the image acquisition module 401 is further configured to control an image acquisition device to perform multi-dimensional shooting on a plurality of circuit board samples, and obtain a plurality of circuit board sample images; and extract an image area containing components of the electricity meter from the plurality of circuit board sample images to obtain a component sample image of the electricity meter.

[0168] In an exemplary embodiment, the apparatus 400 further includes: a model deployment module, configured to export and package the target component recognition model of the electricity meter to obtain a target component recognition model to be deployed; and deploy the target component recognition model to be deployed to a target server.

[0169] Each module in the above-mentioned component recognition device of the electricity meter can be implemented in whole or in part by software, hardware, and their combination. The above-mentioned modules can be embedded in the processor of the computer device in hardware form or be independent of it, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

[0170] In an exemplary embodiment, a computer device is provided. The computer device can be a terminal, and its internal structure diagram can be as Figure 5As shown in the figure. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface, the display unit, and the input device are connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner, and the wireless manner can be achieved through WIFI, a mobile cellular network, NFC (Near Field Communication), or other technologies. The computer program, when executed by the processor, implements a method for identifying components of an electric energy meter. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0171] Those skilled in the art can understand that Figure 5 the structure shown in the figure is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have a different component layout.

[0172] In an exemplary embodiment, a computer device is further provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the steps in the above method embodiments are implemented.

[0173] In an exemplary embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by the processor, the steps in the above method embodiments are implemented.

[0174] In an exemplary embodiment, a computer program product is provided, including a computer program. When the computer program is executed by the processor, the steps in the above method embodiments are implemented.

[0175] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memories. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., without limitation.

[0176] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0177] The above-described embodiments only represent several implementation manners of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A method for identifying components of an electric energy meter, characterized in that: The method comprises: Obtain sample images of electric energy meter components; The sample images of the electric energy meter components are divided into a training verification set to be labeled and a test set, and the training verification set to be labeled is labeled to obtain a target training verification set; the target training verification set includes a target training set and a target verification set; Uploading the target training validation set and the test set to the model training platform; Creating a model training project corresponding to the target training validation set and the test set on the model training platform; Updating the training parameters corresponding to the model training project, and using the updated training parameters and the target training set to train the electric energy meter component recognition model to be trained, to obtain a trained electric energy meter component recognition model; Using the target verification set to verify the trained electric energy meter component identification model, to obtain a verification result of the trained electric energy meter component identification model, and using the test set to test the trained electric energy meter component identification model, to obtain a test result of the trained electric energy meter component identification model; When both the verification result and the test result indicate passing, using the trained electric energy meter component recognition model as a target electric energy meter component recognition model; The target electric energy meter component recognition model is used to perform electric energy meter component recognition on the electric energy meter component image to be recognized, and an electric energy meter component recognition result of the electric energy meter component image to be recognized is obtained.

2. The method according to claim 1, characterized in that The step of using the target electric energy meter component recognition model to perform electric energy meter component recognition on the electric energy meter component image to be recognized, and obtaining an electric energy meter component recognition result of the electric energy meter component image to be recognized, comprises: Acquire the image of the electric energy meter component to be identified; Inputting the to-be-identified electric energy meter component image into the target electric energy meter component identification model for identification, and obtaining category information and position information of the electric energy meter components in the to-be-identified electric energy meter component image; According to the category information and position information of the electric energy meter components, an electric energy meter component recognition result of the electric energy meter component image to be recognized is determined.

3. The method according to claim 1, characterized in that After the target electric energy meter component recognition model is used to perform electric energy meter component recognition on the electric energy meter component image to be recognized and an electric energy meter component recognition result of the electric energy meter component image to be recognized is obtained, the method further includes: Determining the category information and location information of the faulty component in the electric energy meter component image according to the electric energy meter component recognition result of the electric energy meter component image to be recognized; The fault location result of the electric energy meter component image to be identified is determined according to the category information and location information of the faulty component.

4. The method according to claim 1, characterized in that The using the target verification set to verify the trained electric energy meter component identification model to obtain a verification result of the trained electric energy meter component identification model includes: Inputting the target verification set into the trained electric energy meter component recognition model for recognition, and obtaining a recognition result of the target verification set; Determining, according to the target verification set, verification indicator information corresponding to the recognition result of the target verification set; The verification result is determined according to the verification indicator information.

5. The method according to claim 1, characterized in that The step of obtaining a sample image of electric energy meter components includes: Controlling an image acquisition device to perform multi-dimensional photography on a plurality of circuit board samples to obtain a plurality of circuit board sample images; An image region containing electric energy meter components is extracted from the plurality of circuit board sample images to obtain the electric energy meter component sample images.

6. The method according to any one of claims 1 to 5, characterized in that When both the verification result and the test result indicate passing, after using the trained electric energy meter component identification model as the target electric energy meter component identification model, the method further includes: Exporting and packaging the target electric energy meter component identification model to obtain a target electric energy meter component identification model to be deployed; The target electric energy meter component identification model to be deployed is deployed to the target server.

7. An electric energy meter component identification device, characterized in that: The device comprises: An image acquisition module, used to acquire sample images of electric energy meter components; An image division module, used for dividing the sample images of the electric energy meter components into a training verification set to be labeled and a test set, and performing labeling processing on the training verification set to be labeled to obtain a target training verification set; the target training verification set includes a target training set and a target verification set; A target uploading module, used to upload the target training verification set and the test set to the model training platform; A project creation module, used to create a model training project corresponding to the target training verification set and the test set on the model training platform; A model training module, used for updating the training parameters corresponding to the model training project, and using the updated training parameters and the target training set to train the electric energy meter component recognition model to be trained, to obtain the trained electric energy meter component recognition model; A model verification module, used to verify the trained electric energy meter component identification model using the target verification set to obtain a verification result of the trained electric energy meter component identification model, and to test the trained electric energy meter component identification model using the test set to obtain a test result of the trained electric energy meter component identification model; A model determination module, configured to use the trained electric energy meter component identification model as a target electric energy meter component identification model when both the verification result and the test result indicate a pass; The image recognition module is used to perform electric energy meter component recognition on the electric energy meter component image to be recognized by using the target electric energy meter component recognition model, and obtain the electric energy meter component recognition result of the electric energy meter component image to be recognized.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images