CN116563243A - Foreign matter detection method, device, computer equipment and storage medium for power transmission line - Google Patents

Abstract

The application relates to a foreign matter detection method, a device, computer equipment and a storage medium of a power transmission line, wherein a low-illumination image optimization model is trained by acquiring a low-illumination image training set of the power transmission line, and the low-illumination image optimization model can improve each image in the low-illumination image training set to perform optimization processing, so that an acquired enhanced image to be detected is clearer, the enhanced image to be detected is acquired based on the low-illumination image optimization model, the enhanced image to be detected is input into the foreign matter detection model, and a target prediction vector of the foreign matter is acquired. The power transmission line can be monitored in real time, and when foreign objects invade, the target prediction vector of the foreign objects can be accurately obtained.

Description

Foreign matter detection method, device, computer equipment and storage medium for power transmission line

technical field

The present application relates to the technical field of foreign matter detection of power transmission lines, and in particular to a method, device, computer equipment, storage medium and computer program product of foreign matter detection of power transmission lines.

Background technique

Foreign object intrusion of transmission lines refers to some external objects (such as branches, birds, aircraft, etc.) entering the transmission line by mistake, which may cause accidents such as line short circuit, tripping, and equipment damage. These accidents will not only bring economic losses to power supply companies, but also threaten the safety of people's lives and property. Therefore, the study of transmission line foreign object intrusion detection technology has important practical significance and application value for improving the safety, reliability and stability of transmission lines.

The existing foreign object intrusion detection of transmission lines is to set up sensors or cameras around the transmission line, and find and locate any foreign objects through manual or automatic algorithms, such as trees, cables, people, etc. entering the restricted area of the line, and there is a problem of low detection accuracy. .

Contents of the invention

Based on this, it is necessary to provide a foreign object detection method, device, computer equipment, computer-readable storage medium and computer program product of a power transmission line capable of improving the detection accuracy in view of the above technical problems.

In a first aspect, the present application provides a foreign object detection method for a power transmission line. The methods include:

Obtain a low-light image training set of transmission lines;

training a low-illuminance image optimization model based on the low-illuminance image training set, and obtaining an enhanced image to be tested based on the low-illuminance image optimization model;

The enhanced image to be tested is input to the foreign matter detection model to output a target prediction vector of the foreign matter.

In one of the embodiments, the acquisition of the low-illuminance image training set of transmission lines includes:

Obtain original images to be tested corresponding to multiple transmission lines to be tested;

removing noise and background irrelevant to image content in each of the original images to be tested to obtain an initial image set;

Perform data enhancement and contrast enhancement processing on the initial image set to obtain a low-illuminance image data set, and extract the low-illuminance image training set from the low-illuminance image data set.

In one of the embodiments, training the low-illuminance image optimization model based on the low-illuminance image data set includes:

The image features of each image in the low-light image training set are extracted through a convolutional neural network;

The low-illuminance image training set and the image features are input to the Transformer model for training to construct the low-illuminance image optimization model.

In one of the embodiments, extracting the image features of each image in the low-light image training set through the convolutional neural network includes:

Convolving and pooling are performed on each image in the low-illuminance image training set to obtain the image features of each image; the image features are high-level abstract features.

In one of the embodiments, the input of the low-illuminance image training set and the image features of each original image to be tested into the Transformer model to construct the low-illuminance image optimization model includes:

performing feature reconstruction on the image features to obtain a sequence of feature vectors;

The feature vector sequence is input to the multi-head self-attention mechanism layer and the feedforward neural network layer of the Transformer model so that each of the image features interacts with other image features in the feature vector sequence to obtain the result features sequence of vectors;

The decoder of the Transformer model decodes the resulting feature vector sequence to obtain a high-quality image corresponding to each image in the low-illuminance image training set.

In one of the embodiments, obtaining the enhanced image to be tested based on the low-illuminance image optimization model includes:

Adding each image in the low-illuminance image training set and the high-quality image corresponding to each image through a residual connection to obtain an enhanced image to be tested corresponding to each original image to be tested.

In one of the embodiments, the original image to be tested includes two images of different exposure times in the same transmission line to be tested; the training of the low-illuminance image optimization model based on the low-illuminance image data set further includes:

The loss function calculation is performed on the original image to be tested with a long exposure time among the enhanced image to be tested and the two images with different exposure times in the same transmission line to be tested by means of the mean square error method, so as to calculate the loss function according to the loss The function minimizes the difference between the enhanced image under test and the high-quality image corresponding to the enhanced image under test.

In one of the embodiments, the inputting the enhanced image to be tested into the foreign object detection model to output the target prediction vector of the foreign object includes:

performing multi-scale feature pyramid extraction on the enhanced image to be tested to obtain target features;

The target feature is input to the RetinaNet head network to obtain the target prediction vector of the foreign object.

In a second aspect, the present application also provides a foreign object detection device for a power transmission line. The devices include:

The data acquisition module is used to acquire the low-illuminance image training set of the transmission line;

A model training module, configured to train a low-illumination image optimization model based on the low-illumination image training set, and obtain an enhanced image to be tested based on the low-illuminance image optimization model;

The foreign object detection module is used to input the enhanced image to be tested into the foreign object detection model to output the target prediction vector of the foreign object.

In a third aspect, the present application also provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the method steps described in any one of the above embodiments when executing the computer program.

In a fourth aspect, 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 method steps described in any one of the above-mentioned embodiments are implemented.

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 method steps described in any of the foregoing embodiments are implemented.

In the foreign object detection method, device, computer equipment, storage medium and computer program product of the above transmission line, the low illumination image optimization model is trained by obtaining the low illumination image training set of the transmission line, and the low illumination image optimization model can improve low illumination image training. The concentrated images are optimized to make the obtained enhanced image to be tested clearer, and the enhanced image to be tested is obtained based on the low-light image optimization model and the original image to be tested of the transmission line, and the enhanced image to be tested is input to the foreign object detection model, so as to obtain the target prediction vector of the foreign body. The power transmission line can be monitored in real time, and when a foreign object invades, the target prediction vector of the foreign object can be accurately obtained.

Description of drawings

Fig. 1 is a schematic flow chart of a foreign matter detection method for a transmission line in an embodiment;

Fig. 2 is a flow schematic diagram of obtaining a low-illuminance image training set of a power transmission line in one embodiment;

FIG. 3 is a schematic flow diagram of training a low-illuminance image optimization model based on a low-illuminance image data set in an embodiment;

Fig. 4 is a schematic flow diagram of constructing a low-illuminance image training optimization model in an embodiment;

Fig. 5 is a schematic flow chart of outputting a target prediction vector of a foreign object in an embodiment;

Fig. 6 is a schematic flow chart of a foreign object detection method for a power transmission line in another embodiment;

Fig. 7 is a structural block diagram of a foreign object detection device for a power transmission line in an embodiment;

Figure 8 is a diagram of the internal structure of a computer device in one embodiment.

Detailed ways

In order to make the purpose, technical solution and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present application, and are not intended to limit the present application.

In one embodiment, as shown in FIG. 1 , a schematic flowchart of a method for detecting foreign matter on a power transmission line, the present application provides a method for detecting foreign matter on a power transmission line, which includes the following steps:

Step 102, acquiring a training set of low-illuminance images of transmission lines.

Among them, the low-illumination image training set is the image after preprocessing the original image to be tested.

Step 104, train a low-illumination image optimization model based on the low-illuminance image training set, and obtain an enhanced image to be tested based on the low-illuminance image optimization model.

Among them, training the low-illumination image optimization model based on the low-illumination image training set can be understood as a process of optimizing training on the images in the low-illumination image training set, and the entire optimization training process constitutes the low-illumination image optimization model. The enhanced image to be tested refers to an image after optimization processing is performed on each image in the low-illuminance image training set.

Step 106, input the enhanced image to be tested into the foreign object detection model to output the target prediction vector of the foreign object.

Wherein, the target prediction vector includes the target location of the foreign object and the category of the foreign object. Specifically, the size of the target prediction vector may represent the target location of the foreign object, and the direction of the target prediction vector may represent the category of the foreign object.

In the foreign object detection method of the above transmission line, the low illumination image optimization model is trained by obtaining the low illumination image training set of the transmission line, and the low illumination image optimization model can improve each image in the low illumination image training set for optimization processing, so that the acquired The enhanced image is clearer, and the enhanced image to be tested is obtained based on the low-light image optimization model and the original image to be tested of the transmission line, and the enhanced image to be tested is input into the foreign object detection model to obtain the target prediction vector of the foreign object. The power transmission line can be monitored in real time, and when a foreign object invades, the target prediction vector of the foreign object can be accurately obtained.

In one embodiment, as shown in FIG. 2 , as shown in FIG. 2 , a schematic flow diagram of obtaining a training set of low-illuminance images of a power transmission line, obtaining a training set of low-illuminance images of a power transmission line includes:

Step 202, acquiring original images to be tested corresponding to a plurality of power transmission lines to be tested.

Wherein, the original image to be tested includes various scene pictures of the power transmission line to be tested taken by a variety of different shooting devices under low illumination conditions. At least two original images to be tested with different exposure times are taken for each transmission line to be tested. At the same time, the original images to be tested with long exposure times need to be preprocessed to reduce noise and increase contrast. Preprocessing methods include histogram equalization , Gaussian blur, median filter, brightness enhancement, etc.

Step 204, removing noise and background irrelevant to image content in each original image to be tested to obtain an initial image set.

Specifically, after the original image to be tested is acquired, the data of the original image to be tested is cleaned to ensure that each original image to be tested is valid without any noise or damage. During the process, various image editing tools can be used to process and repair to remove the background and noise irrelevant to the image content.

Step 206, perform data enhancement and contrast enhancement processing on the initial image set to obtain a low-illuminance image data set, and extract a low-illuminance image training set from the low-illuminance image data set.

Among them, data enhancement techniques include cropping, rotating, scaling, translating, adding noise, etc.

In this embodiment, for obtaining the original images to be tested corresponding to a plurality of power transmission lines to be detected, the noise and background irrelevant to the image content in the original image data to be tested are removed to obtain an initial image set, and further, the initial image Data enhancement and contrast enhancement processing are performed on the set, so as to increase the amount of image data and enhance the generalization ability of the subsequent low-light image optimization model trained based on the low-light image training set. Therefore, the image after the above processing (that is, the low-illuminance image dataset) can obtain images with higher definition, and the low-illuminance image optimization model trained by the low-illuminance image training set extracted from the low-illuminance image dataset can obtain Higher quality images.

In one embodiment, as shown in FIG. 3 , a schematic flow chart of training a low-illuminance image optimization model based on a low-illumination image dataset, training a low-illumination image optimization model based on a low-illumination image dataset includes:

Step 302, extracting the image features of each image in the training set of low-illuminance images through a convolutional neural network.

Among them, Convolutional Neural Networks (CNN) is a kind of Feedforward Neural Networks (Feedforward Neural Networks) that includes convolution calculation and has a deep structure, and is one of the representative algorithms of deep learning. The convolutional neural network has the ability of representation learning, and can perform shift-invariant classification on the input information according to its hierarchical structure. A convolutional neural network consists of an input layer, a pooling layer, and an output layer. Among them, the output layer uses a logistic function or a normalized exponential function (softmax function) to output classification labels. Specifically, in this embodiment, Resnet50 (Residual Network, residual network) can be used to realize the above functions, and the image features of each image in the training set of low-illuminance images can be extracted through Resnet50.

Step 304, input the low-illuminance image training set and image features into the Transformer model for training, so as to build a low-illuminance image optimization model.

In this embodiment, the convolutional neural network is used to extract the image features of each image in the low-light image training set. Based on the image features of each image and the low-light image training set, the Transformer model can be used to train low-light image optimization that can enhance image clarity. Model.

In one embodiment, extracting the image features of each image in the low-illuminance image training set through a convolutional neural network includes: performing convolution and pooling on each image in the low-illuminance image training set to obtain the image features of each image; the image features are High-level abstract features.

In this embodiment, specifically, the convolutional neural network may be ResNet50, and ResNet50 is used to perform convolution and pooling processing on each image in the low-illuminance image training set to obtain high-level abstract features of each image, and obtain a fixed-size Feature map (i.e., image features), where the fixed size refers to the size of each image in the low-light image training set after sampling at a fixed magnification. The acquired image features are helpful for further feature reconstruction of the image to obtain a higher-definition image.

In one embodiment, as shown in Figure 4, a schematic flow diagram of constructing a low-illuminance image training optimization model; the low-illuminance image training set and the image features of each original image to be tested are input to the Transformer model to construct a low-illuminance image optimization model include:

Step 402, performing feature reconstruction on image features to obtain feature vector sequences.

Specifically, the image features are expanded into a sequence of feature vectors by row or column, each vector represents a row or column in the feature map, and has a spatial sequence relationship.

Step 404, input the feature vector sequence to the multi-head self-attention mechanism layer and the feed-forward neural network layer of the Transformer model so that each image feature interacts with other image features in the feature vector sequence to obtain the resulting feature vector sequence.

Specifically, the encoder of the Transformer model is used to encode the sequence of feature vectors. In the encoder of the Transformer model, each feature vector is fed into a multi-head self-attention mechanism layer and a feed-forward neural network layer for processing, and each vector interacts with other vectors in the sequence.

Step 406, the decoder of the Transformer model decodes the resulting feature vector sequence to obtain high-quality images corresponding to each image in the low-illuminance image training set.

In this embodiment, image features are reconstructed to obtain a feature vector sequence, and the feature vector sequence is input to the multi-head self-attention mechanism layer and the feed-forward neural network layer of the Transformer model so that each image feature is consistent with the feature vector Interact with other image features in the sequence to obtain the result feature vector sequence. Since each vector will interact with other vectors in the sequence, richer context information can be obtained, and then the result feature vector can be analyzed by the decoder of the Transformer model Sequences are decoded, and the high-quality images corresponding to each image in the obtained low-light image training set have higher definition.

In one embodiment, obtaining the enhanced image to be tested based on the low-illuminance image optimization model includes: adding each image in the low-illuminance image training set and the high-quality image corresponding to each image through a residual connection to obtain the corresponding value of each original image to be tested. The enhanced image to be tested.

In this embodiment, each image in the training set of low-illumination images and the high-quality image corresponding to each image are added through the residual connection, so the obtained enhanced image corresponding to each original image to be tested is compared with the high-quality image The image has a higher similarity with the original image to be tested and is clearer.

In one embodiment, the original image to be tested includes two images of different exposure times in the same transmission line to be tested; training the low-illuminance image optimization model based on the low-illuminance image data set also includes: enhancing the test by the mean square error method The loss function calculation is performed on the original image to be tested with a long exposure time in the image and the two images of different exposure times in the same power transmission line to be tested, so as to minimize the corresponding enhanced image to be tested and the enhanced image to be tested according to the loss function The difference between high-quality images.

Among them, the mean-square error (mean-square error, MSE) is a measure that reflects the degree of difference between the estimator and the estimated quantity.

In this embodiment, the loss function calculation is performed on the original image to be tested with a long exposure time among the enhanced image to be tested and the two images with different exposure times in the same transmission line to be tested by means of the mean square error method, which can be based on the acquired The loss function optimizes the low-light image optimization model to reduce the difference between the enhanced image to be tested and the high-quality image corresponding to the enhanced image to be tested, so that the final enhanced image to be tested is closer to the corresponding high-quality image The image, which is closer to the original image to be tested, provides a more accurate detection scene for the subsequent foreign object detection of the transmission line based on the enhanced image to be tested.

In one embodiment, a low-illumination image verification set and a low-illumination image test set can also be extracted based on the low-illumination image data set; the low-illumination image verification set is used to adjust the hyperparameters of the low-illumination image optimization model, and the low-illumination image test set is used for Evaluate the performance of optimized models for low-light images.

In this embodiment, a low-illuminance image optimization model with a higher optimization effect can be obtained through the low-illuminance image verification set and the low-illuminance image test set.

In one embodiment, a schematic flow chart of outputting a target prediction vector of a foreign object as shown in FIG. 5 ; the enhanced image to be tested is input into a foreign object detection model to output a target prediction vector of a foreign object, including:

Step 502, performing multi-scale feature pyramid extraction on the enhanced image to be tested to obtain target features.

Among them, the multi-scale feature pyramid can be divided into Gaussian pyramid and Laplacian pyramid. Among them, the Gaussian pyramid is a series of images obtained by downsampling the maximum resolution image at the bottom successively. The resolution of the bottom image is the highest, and the resolution of the image is lower as it goes up; the Laplacian pyramid can be considered as a residual pyramid , which is used to store the difference between the downsampled image and the original image.

The backbone network of the foreign object detection model has the same structure as the low-light image optimization model, but the specific parameters are different.

Step 504, input the target feature into the RetinaNet head network to obtain the target prediction vector of the foreign object.

In this embodiment, the foreign object detection model performs multi-scale feature pyramid extraction on the enhanced image to be tested, and inputs the extracted target features to the RetinaNet head network, thereby outputting the target prediction vector of the detected foreign object, specifically, the foreign object The size of the target prediction vector represents the target location (coordinates) of the foreign object, and the direction of the target prediction vector of the foreign object represents the category of the foreign object. Therefore, based on the above method, it is possible to perform accurate foreign object detection on the transmission line to be tested, and output the target location and category of the foreign object.

In one embodiment, in the process of training the foreign object detection model, the gap between the target prediction vector and the true value vector can also be measured through the loss function (focalloss) and the segmentation function (smooth-l1 loss), and can be compared with the image enhancement The loss function together as the total loss for model optimization. Among them, the truth vector refers to the information of the target detection position in the low-light image data set. When training the foreign object detection model, the information of the target detection position will exist in the form of artificial labels. At the same time, the parameters of the foreign object detection model are optimized using the backpropagation algorithm to minimize the value of the loss function. In order to avoid overfitting, it is generally necessary to use some regularization techniques, such as random deactivation (dropout) or L2 regularization. Therefore, the foreign object detection model can detect foreign objects on the transmission line more accurately.

In one embodiment, as shown in FIG. 6 , another schematic flow chart of a foreign object detection method for a power transmission line; the foreign object detection method for a power transmission line includes:

Step 602, acquiring original images to be tested corresponding to a plurality of power transmission lines to be tested.

Step 604, removing noise and background irrelevant to image content in each original image to be tested to obtain an initial image set.

Step 606: Perform data enhancement and contrast enhancement processing on the initial image set to obtain a low-illuminance image dataset, and extract a low-illuminance image training set from the low-illuminance image dataset.

Step 608, perform convolution and pooling on each image in the low-illuminance image training set to obtain image features of each image; the image features are high-level abstract features.

Step 610, performing feature reconstruction on image features to obtain feature vector sequences.

Step 612, input the feature vector sequence to the multi-head self-attention mechanism layer and the feed-forward neural network layer of the Transformer model so that each image feature interacts with other image features in the feature vector sequence to obtain the resulting feature vector sequence.

In step 614, the decoder of the Transformer model decodes the resulting feature vector sequence to obtain high-quality images corresponding to each image in the low-illuminance image training set.

Step 616: Add the images in the low-illuminance image training set and the high-quality images corresponding to the images through the residual connection to obtain the enhanced images to be tested corresponding to the original images to be tested.

Step 618, performing multi-scale feature pyramid extraction on the enhanced image to be tested to obtain target features.

Step 620, input the target feature into the RetinaNet head network to obtain the target prediction vector of the foreign object.

In this embodiment, the original images to be tested corresponding to a plurality of transmission lines to be tested are obtained, and the original images to be tested are subjected to preliminary processing: noise and background irrelevant to the image content in each original image to be tested are removed to obtain the initial image set; perform data enhancement and contrast enhancement processing on the initial image set to obtain a low-light image data set, and extract a low-light image training set from the low-light image data set. Therefore, the clarity and brightness of the original image to be tested can be improved through the above processing, making the picture captured by the surveillance camera brighter and clearer, thereby reducing missed detection and improving detection efficiency. Further, by performing convolution and pooling on each image in the low-light image training set to obtain the image features of each image, perform feature reconstruction on the image features to obtain a feature vector sequence, and input the feature vector sequence to the multi-head Transformer model The self-attention mechanism layer and the feed-forward neural network layer allow each image feature to interact with other image features in the feature vector sequence to obtain the result feature vector sequence, and the decoder of the Transformer model decodes the result feature vector sequence to obtain Obtain the high-quality images corresponding to each image in the low-light image training set, and add the high-quality images corresponding to each image in the low-light image training set and each image through residual connection to obtain the enhanced image to be tested corresponding to each original image to be tested , through the above method, the contrast and details of each image in the low-light image training set can be improved, so that the target feature obtained by the multi-scale feature pyramid extraction based on the enhanced image to be tested and the target prediction vector of the foreign object obtained based on the target feature are more accurate. , so as to reduce the misjudgment of foreign matter detection and improve the detection accuracy.

It should be understood that although the steps in the flow charts involved in the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed sequentially in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least some of the steps in the flow charts involved in the above embodiments may include multiple steps or stages, and these steps or stages are not necessarily executed at the same time, but may be executed at different times, The execution order of these steps or stages is not necessarily performed sequentially, but may be executed in turn or alternately with other steps or at least a part of steps or stages in other steps.

Based on the same inventive concept, an embodiment of the present application further provides a foreign object detection device for a transmission line for realizing the above-mentioned foreign object detection method for a transmission line. The solution to the problem provided by the device is similar to the implementation described in the above method, so the specific limitations in the embodiments of the foreign object detection device for one or more power transmission lines provided below can be referred to above for power transmission lines The limitation of the foreign object detection method will not be repeated here.

In one embodiment, as shown in FIG. 7 , a structural block diagram of a foreign object detection device for a power transmission line, the foreign object detection device 700 provided by the present application includes: a data acquisition module 710, a model training module 720, and a foreign object detection module 730, wherein:

The data acquisition module 710 is configured to acquire a training set of low-illuminance images of power transmission lines.

The model training module 720 is configured to train a low-illumination image optimization model based on the low-illuminance image training set, and obtain an enhanced image to be tested based on the low-illuminance image optimization model.

The foreign object detection module 730 is configured to input the enhanced image to be tested into the foreign object detection model to output a target prediction vector of the foreign object.

In one embodiment, the data acquisition module is also used to acquire the original images to be tested corresponding to a plurality of transmission lines to be tested; remove the noise and background irrelevant to the image content in each original image to be tested to obtain the initial image set; Perform data enhancement and contrast enhancement processing on the initial image set to obtain a low-illumination image dataset, and extract a low-illumination image training set from the low-illumination image dataset.

In one embodiment, the model training module is also used to extract the image features of each image in the low-illuminance image training set through a convolutional neural network; the low-illuminance image training set and the image features are input to the Transformer model for training to construct low-illuminance images Optimize the model.

In one embodiment, the model training module is further configured to perform convolution and pooling on each image in the low-illuminance image training set to obtain image features of each image; the image features are high-level abstract features.

In one embodiment, the model training module is also used to perform feature reconstruction on the image features to obtain the feature vector sequence; the feature vector sequence is input to the multi-head self-attention mechanism layer and the feedforward neural network layer of the Transformer model to make each image The feature interacts with other image features in the feature vector sequence to obtain the result feature vector sequence; the decoder of the Transformer model decodes the result feature vector sequence to obtain high-quality images corresponding to each image in the low-light image training set.

In one embodiment, the model training module is further configured to add each image in the low-illuminance image training set and the high-quality image corresponding to each image through a residual connection to obtain an enhanced image to be tested corresponding to each original image to be tested.

In one embodiment, the original image to be tested includes two images of different exposure times in the same transmission line to be tested; the model training module is also used in the enhanced image to be tested and the same transmission line to be tested by the mean square error method Among the two images with different exposure times, the loss function is calculated for the original image to be tested with a long exposure time, so as to minimize the difference between the enhanced image to be tested and the high-quality image corresponding to the enhanced image to be tested according to the loss function.

In one embodiment, the foreign object detection module is used to perform multi-scale feature pyramid extraction on the enhanced image to be tested to obtain target features; input the target features to the RetinaNet head network to obtain a target prediction vector of foreign objects.

Each module in the above-mentioned foreign object detection device for transmission lines can be fully or partially realized by software, hardware and combinations thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, and can also be stored in the memory of the computer device in the form of software, so that the processor can invoke and execute the corresponding operations of the above-mentioned modules.

In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure may be as shown in FIG. 8 . The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Wherein, the processor of the computer device is used to provide calculation 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 computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be realized through WIFI, mobile cellular network, NFC (Near Field Communication) or other technologies. When the computer program is executed by a processor, a method for detecting foreign matter of a power transmission line is realized. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a button, a trackball or a touch pad provided on the casing of the computer device , and can also be an external keyboard, touchpad, or mouse.

Those skilled in the art can understand that the structure shown in FIG. 8 is only a block diagram of a partial structure related to the solution of this application, and does not constitute a limitation on the computer equipment to which the solution of this application is applied. The specific computer equipment can be More or fewer components than shown in the figures may be included, or some components may be combined, or have a different arrangement of components.

In one embodiment, there is also provided a computer device, including a memory and a processor, where a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps in the foregoing method embodiments are implemented.

In one embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the foregoing method embodiments are implemented.

Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above-mentioned embodiments can be completed by instructing related hardware through computer programs, and the computer programs can be stored in a non-volatile computer-readable memory In the medium, when the computer program is executed, it may include the processes of the embodiments of the above-mentioned methods. Wherein, any reference to storage, database or other media used in the various embodiments provided in the present application may include at least one of non-volatile and volatile storage. Non-volatile memory can include read-only memory (Read-Only Memory, ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive variable memory (ReRAM), magnetic variable memory (Magnetoresistive Random Access Memory, MRAM), Ferroelectric Random Access Memory (FRAM), Phase Change Memory (Phase Change Memory, PCM), graphene memory, etc. The volatile memory may include random access memory (Random Access Memory, RAM) or external cache memory. As an illustration and not a limitation, RAM can be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (Dynamic Random Access Memory, DRAM). The databases involved in the various embodiments provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a blockchain-based distributed database, etc., but is not limited thereto. The processors involved in the various embodiments provided by this application can be general-purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices, data processing logic devices based on quantum computing, etc., and are not limited to this.

The technical features of the above embodiments can be combined arbitrarily. To make the description concise, all possible combinations of the technical features in the above embodiments are not described. However, as long as there is no contradiction in the combination of these technical features, they should be It is considered to be within the range described in this specification.

The above-mentioned embodiments only express several implementation modes of the present application, and the description thereof is relatively specific and detailed, but should not be construed as limiting the patent scope of the present application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be determined by the appended claims.

Claims (12)

1. A foreign matter detection method of a power transmission line, characterized in that the method comprises: Obtain a low-light image training set of transmission lines; training a low-illuminance image optimization model based on the low-illuminance image training set, and obtaining an enhanced image to be tested based on the low-illuminance image optimization model; The enhanced image to be tested is input to the foreign matter detection model to output a target prediction vector of the foreign matter. 2. The foreign object detection method of transmission line according to claim 1, is characterized in that, the low-illuminance image training set of described acquisition transmission line comprises: Obtain original images to be tested corresponding to multiple transmission lines to be tested; removing noise and background irrelevant to image content in each of the original images to be tested to obtain an initial image set; Perform data enhancement and contrast enhancement processing on the initial image set to obtain a low-illuminance image data set, and extract the low-illuminance image training set from the low-illuminance image data set. 3. The foreign object detection method of transmission line according to claim 1, is characterized in that, based on described low-illuminance image dataset training low-illuminance image optimization model comprises: The image features of each image in the low-light image training set are extracted through a convolutional neural network; The low-illuminance image training set and the image features are input to the Transformer model for training to construct the low-illuminance image optimization model. 4. the foreign object detection method of power transmission line according to claim 3, is characterized in that, extracting the image feature of each image in low-illuminance image training set by convolutional neural network comprises: Convolving and pooling are performed on each image in the low-illuminance image training set to obtain the image features of each image; the image features are high-level abstract features. 5. The foreign matter detection method of power transmission line according to claim 3, characterized in that, the described low-illuminance image training set and the image features of each original image to be tested are input to the Transformer model to construct the Low-light image optimization models include: performing feature reconstruction on the image features to obtain a sequence of feature vectors; The feature vector sequence is input to the multi-head self-attention mechanism layer and the feedforward neural network layer of the Transformer model so that each of the image features interacts with other image features in the feature vector sequence to obtain the result features sequence of vectors; The decoder of the Transformer model decodes the resulting feature vector sequence to obtain a high-quality image corresponding to each image in the low-illuminance image training set. 6. The foreign matter detection method of a power transmission line according to claim 5, wherein obtaining the enhanced image to be tested based on the low-illuminance image optimization model comprises: Adding each image in the low-illuminance image training set and the high-quality image corresponding to each image through a residual connection to obtain an enhanced image to be tested corresponding to each original image to be tested. 7. The foreign matter detection method of a power transmission line according to claim 5, wherein the original image to be tested includes two images of different exposure times in the same power transmission line to be detected; The dataset for training low-light image optimization models also includes: The loss function calculation is performed on the original image to be tested with a long exposure time among the enhanced image to be tested and the two images with different exposure times in the same transmission line to be tested by means of the mean square error method, so as to calculate the loss function according to the loss The function minimizes the difference between the enhanced image under test and the high-quality image corresponding to the enhanced image under test. 8. The foreign matter detection method of a power transmission line according to claim 1, wherein said inputting the enhanced image to be tested into a foreign matter detection model to output a foreign matter target prediction vector comprises: performing multi-scale feature pyramid extraction on the enhanced image to be tested to obtain target features; The target feature is input to the RetinaNet head network to obtain the target prediction vector of the foreign object. 9. A foreign object detection device for a transmission line, characterized in that the device comprises: The data acquisition module is used to acquire the low-illuminance image training set of the transmission line; A model training module, configured to train a low-illumination image optimization model based on the low-illumination image training set, and obtain an enhanced image to be tested based on the low-illuminance image optimization model; The foreign object detection module is used to input the enhanced image to be tested into the foreign object detection model to output the target prediction vector of the foreign object. 10. A computer device, comprising a memory and a processor, the memory stores a computer program, wherein the processor implements the method according to any one of claims 1 to 8 when executing the computer program step. 11. A computer-readable storage medium, on which a computer program is stored, wherein, when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are realized. 12. 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 8 are realized.