Power transmission line hidden danger target detection method and device, electronic equipment and storage medium
Through the pre-trained YOLOv5s deep learning model, combined with the self-attention mechanism and FReLU nonlinear activation function, hidden danger targets in transmission line images are automatically identified, solving the problems of low efficiency and insufficient accuracy of manual detection, realizing efficient and accurate hidden danger detection, and ensuring the safety of transmission lines.
Patent Information
- Application Number
- CN202411245914.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-06
- Publication Date
- 2025-09-12
AI Technical Summary
In the existing technology, the detection of hidden dangers in transmission lines relies on manual inspection, which leads to waste of resources and missed detections and false detections. It is impossible to maintain efficient monitoring for a long time and there are safety hazards.
The pre-trained YOLOv5s deep learning model is used, combined with the self-attention mechanism and the FReLU nonlinear activation function, to automatically identify hidden danger targets in transmission line images. The feature map is reconstructed through the self-attention mechanism to enhance the spatial sensitivity and generalization ability of the model.
It improves the accuracy and efficiency of hidden danger detection of transmission lines, reduces manual intervention, and ensures the safe operation of transmission lines.
Smart Images
Figure CN120635398A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of power supply technology, and in particular to a method, device, electronic device and storage medium for detecting hidden danger targets in power transmission lines. Background Art
[0002] As a key component of the power system, the safe operation of transmission lines is of paramount importance. However, transmission lines often face various hidden dangers such as aging, external damage, and natural damage. If these hidden dangers are not discovered and eliminated in a timely manner, they will lead to serious power outages and property losses.
[0003] Currently, relevant technologies require manual inspection and screening of transmission line hidden danger images, but a large number of transmission line images without warning will waste huge human and material resources. At the same time, manual labor cannot maintain the energy for long-term monitoring, resulting in the disadvantages of missed detection and false detection. Summary of the Invention
[0004] In view of the above-mentioned defects or deficiencies in the relevant technology, it is desired to provide a method, device, electronic equipment and storage medium for detecting hidden danger targets in transmission lines, which can efficiently identify hidden danger targets, improve accuracy and ensure the safe operation of transmission lines.
[0005] In a first aspect, the present application provides a method for detecting hidden danger targets in power transmission lines, the method comprising:
[0006] Acquire a transmission line image sequence;
[0007] The transmission line image sequence is input into a pre-trained deep learning model to identify transmission line images with hidden danger targets. The deep learning model is a YOLOv5s model. A self-attention mechanism unit is arranged between the Backbone network and the Neck network in the YOLOv5s model. The YOLOv5s model adopts the FReLU nonlinear activation function, which is obtained by expanding the ReLU function and the PReLU function into a two-dimensional activation function.
[0008] Optionally, in some embodiments of the present application, the self-attention mechanism unit includes a channel attention subunit and a spatial attention subunit, the channel attention subunit is configured to process the input feature map to obtain a channel attention map, and the spatial attention subunit is configured to obtain an output feature map based on the input feature map and the channel attention map.
[0009] Optionally, in some embodiments of the present application, the channel attention subunit is specifically used to input the input feature map into the maximum pooling channel and the average pooling channel respectively, and generate an intermediate vector; obtain a feature vector by passing the intermediate vector through a multi-layer perceptron, and perform item-by-item multiplication and nonlinear activation on the feature vector to obtain the channel attention map.
[0010] Optionally, in some embodiments of the present application, the spatial attention sub-unit is specifically used to process the multiplication result to obtain a spatial attention map, and the operation object of the multiplication result is the input feature map and the channel attention map; and, the spatial attention map is multiplied with the multiplication result again to obtain the output feature map.
[0011] Optionally, in some embodiments of the present application, the spatial attention subunit is further specifically used to perform maximum pooling and average pooling operations on the multiplication results to generate an intermediate feature map; and perform convolution operations and nonlinear activation on the intermediate feature map to obtain the spatial attention map.
[0012] Optionally, in some embodiments of the present application, the activation threshold of the FReLU nonlinear activation function is a funnel feature map generated by performing a two-dimensional convolution operation on the input feature map.
[0013] Optionally, in some embodiments of the present application, the deep learning model is trained by the following steps:
[0014] Obtain hidden danger images of transmission lines at different voltage levels;
[0015] Labeling the transmission line hidden danger images by type and dividing them into training sets and test sets;
[0016] The deep learning architecture is trained using the training set, and the training results are verified and weighted using the test set to obtain the deep learning model.
[0017] In a second aspect, the present application provides a device for detecting hidden danger targets in power transmission lines, the device comprising:
[0018] an acquisition module configured to acquire a sequence of transmission line images;
[0019] The recognition module is configured to input the transmission line image sequence into a pre-trained deep learning model to identify transmission line images with hidden danger targets, wherein the deep learning model is a YOLOv5s model, a self-attention mechanism unit is provided between the Backbone network and the Neck network in the YOLOv5s model, and the YOLOv5s model adopts the FReLU nonlinear activation function, which is obtained by expanding the ReLU function and the PReLU function into a two-dimensional activation function.
[0020] In a third aspect, the present application provides an electronic device, comprising a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the instruction, the program, the code set or the instruction set is loaded and executed by the processor to implement the steps of the transmission line hidden danger target detection method described in any one of the first aspects.
[0021] In a fourth aspect, the present application provides a computer-readable storage medium, which stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of the power transmission line hidden danger target detection method described in any one of the first aspects.
[0022] It can be seen from the above technical solutions that the embodiments of the present application have the following advantages:
[0023] The embodiments of the present application provide a method, device, electronic device and storage medium for detecting hidden danger targets in transmission lines. The method automatically identifies transmission line images with hidden danger targets in a transmission line image sequence through a pre-trained deep learning model, without the need for manual detection, and is efficient and convenient. The deep learning model can use the self-attention mechanism to reconstruct the feature maps in the network, highlight important features, and improve the model's generalization ability for different scenarios. The FReLU nonlinear activation function is used to enhance the model's spatial sensitivity, so that ordinary convolution operations can also capture complex visual layouts, while giving the model pixel-level modeling capabilities, thereby greatly improving recognition accuracy and ensuring the safe operation of transmission lines. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0025] Figure 1A flowchart of a method for detecting hidden danger targets in power transmission lines provided in an embodiment of the present application;
[0026] Figure 2 A schematic diagram of the YOLOv5s network structure provided in an embodiment of the present application;
[0027] Figure 3 A schematic diagram of the structure of a self-attention mechanism unit provided in an embodiment of the present application;
[0028] Figure 4 A schematic diagram of a processing flow of a channel attention sub-unit provided in an embodiment of the present application;
[0029] Figure 5 A schematic diagram of a processing flow of a spatial attention subunit provided in an embodiment of the present application;
[0030] Figure 6 A schematic diagram of the principle of a FReLU nonlinear activation function provided in an embodiment of the present application;
[0031] Figure 7 This is a structural block diagram of a device for detecting hidden dangers in power transmission lines provided in an embodiment of the present application;
[0032] Figure 8 This is a structural block diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0033] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.
[0034] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.
[0035] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of this application can be combined with each other. Figures 1 to 8 The present invention provides a detailed description of the method, device, electronic device and storage medium for detecting hidden danger targets in power transmission lines provided by the embodiments of the present application.
[0036] Please refer to Figure 1 , which is a flow chart of a method for detecting hidden danger targets in power transmission lines provided in an embodiment of the present application, the method specifically comprises the following steps:
[0037] S101: Acquire a transmission line image sequence.
[0038] For example, in the embodiments of the present application, transmission line image sequences can be acquired in real time using monitoring equipment such as cameras, thereby ensuring continuous data collection without omissions. Alternatively, transmission line image sequences can be acquired periodically by monitoring equipment, for example, every two minutes, thereby extending the life of the monitoring equipment and reducing the amount of data processing. Alternatively, transmission line image sequences can be acquired with longer acquisition intervals during the day and shorter acquisition intervals at night, providing greater flexibility.
[0039] S102: Input the transmission line image sequence into a pre-trained deep learning model to identify transmission line images with hidden danger targets. The deep learning model is a YOLOv5s model. A self-attention mechanism unit is set between the Backbone network and the Neck network in the YOLOv5s model. The YOLOv5s model uses the FReLU nonlinear activation function, which is obtained by expanding the ReLU function and the PReLU function into a two-dimensional activation function.
[0040] For example, ① Figure 2 As shown in FIG, it is a schematic diagram of a YOLOv5s network structure provided by an embodiment of the present application. Among them, the Focus unit can slice the image; the CBL unit can perform operations such as convolution (Conv), batch normalization (BN) and activation (Leaky relu) on the image; and the CSP unit can divide the feature map of the base layer into two parts and merge them through a cross-stage hierarchical structure, thereby reducing the amount of calculation and improving the accuracy of the algorithm. Further, as Figure 3 As shown in the figure, the self-attention mechanism unit (Convolutional Block Attention Module, CBAM) set between the Backbone network and the Neck network can include a channel attention subunit CAM and a spatial attention subunit SAM. The channel attention subunit CAM can process the input feature map to obtain a channel attention map, and the spatial attention subunit SAM can obtain an output feature map based on the input feature map and the channel attention map. That is, by setting the self-attention mechanism unit, the channel attention and spatial attention can be combined to achieve adaptive feature refinement of the model, thereby enhancing the target feature extraction capability of the YOLOv5s network. For example Figure 4As shown in the figure, the channel attention subunit CAM is specifically used to input the input feature map into the maximum pooling channel and the average pooling channel respectively, and generate an intermediate vector, thereby aggregating the spatial information of the target feature map and reducing the loss of target information. The intermediate vector is then passed through a multi-layer perceptron MLP containing a single hidden layer to obtain the feature vector, which reduces the calculation level of the unit. The feature vector is then multiplied item by item and nonlinear activations such as Sigmoid are performed to obtain the channel attention map, thereby extracting the contour information of the target.
[0041] For another example, the spatial attention subunit SAM is specifically used to process the multiplication result to obtain a spatial attention map. The operation object of the multiplication result is the input feature map and the channel attention map. The spatial attention map is then multiplied with the multiplication result again to obtain an output feature map, thereby extracting the spatial position information of the target feature map and supplementing the channel attention map. Figure 5 As shown in the figure, the spatial attention subunit SAM is also specifically used to perform maximum pooling and average pooling operations on the multiplication results to generate an intermediate feature map, and then perform convolution operations and nonlinear activations such as Sigmoid on the intermediate feature map to obtain a spatial attention map.
[0042] ② For example, the activation threshold of the FReLU nonlinear activation function is a funnel feature map generated by performing a two-dimensional convolution operation on the input feature map. Specifically, function ReLU() = max(0,x), function PReLU(x) = max(0,x) + αmin(0,x), where α is a learnable parameter. The FReLU nonlinear activation function extends the functions ReLU and PReLU to two-dimensional activation functions by processing the input feature map in the spatial dimension. Furthermore, the embodiment of the present application can perform a two-dimensional convolution operation on the input feature map to extract the spatial pattern in the input feature map, and generate a funnel feature map for adjusting the activation response to replace the zero threshold in the function ReLU and the function PreLU. This activation threshold is spatially variable, allowing more complex nonlinear responses, so that the activation threshold is applied to the input feature map, and the degree of activation of each position is determined according to the new value. For example, the FReLU nonlinear activation function can be y=max(x,Conv2D(x)), where max() represents the maximum value operation, which is similar to the nonlinear activation form of the standard ReLU function, x represents the input feature map, and Conv2D() represents a two-dimensional convolution operation, which is used to extract local patterns in the input feature map.
[0043] For example Figure 6As shown, the FReLU nonlinear activation function uses a 3×3 deep convolution kernel to capture spatial information and integrate local feature patterns. That is, when the input feature map x passes through the FReLU nonlinear activation function, it retains both the original input and the result of the convolution. By taking the maximum value, it not only retains the main information in the input feature map, but also integrates the local patterns extracted from the convolution operation. In short, the FReLU nonlinear activation function extends the ReLU and PReLU functions into two-dimensional activation functions by introducing negligible spatial conditional overhead. It can achieve pixel-level spatial information modeling capabilities in the activation function stage, making it simple and efficient in visual recognition tasks such as target detection and semantic segmentation. At the same time, it has demonstrated significant improvements and robustness in experiments on multiple datasets and visual tasks, proving that it can effectively improve model performance and adapt to different scenarios.
[0044] ③ In the process of training the deep learning model, the embodiment of the present application first obtains the hidden danger images of the transmission lines of different voltage levels. For example, a total of 23,780 hidden danger images are collected through the transmission line visualization platform of each provincial grid company. The hidden danger types include but are not limited to cranes, tower cranes, bulldozers, forklifts, muck trucks, excavators, kites, balloons, plastic films, dust nets, reflective films, wildfires and smoke, etc., and fully considers the hidden danger type characteristics, color, angle, occlusion rate and acquisition time. Factors such as; then, the transmission line hidden danger image is type-labeled and labeled as Y The OLO dataset format is used and divided into training and test sets. For example, 23,780 images of transmission line hidden dangers are annotated, and 2,378 images of nine major hidden danger types are selected as the test set, and the remaining 21,402 images are used as the training set. Then, the deep learning architecture is trained using the training set, and the training results are verified and weighted using the test set to obtain a deep learning model. For example, in the Python 3.8 environment, experiments are conducted using the Si Teng Heli IT4200-2GLLH workstation, which is equipped with an NVIDIA TITAN RTX graphics card (16GB video memory) and 1TB system memory, running Ubuntu 20.04 system, and configured with CUDA 10.0 and CUDNN v7.6.5 acceleration environment, which provides strong support for the improved YOLOv5s model. The image input size selected for training is 640×640, the batch training size is 16, the learning rate is set to 0.01, and the training step size is set to 200. Through model training and result analysis, compared with the original YOLOv5s model, the improved YOLOv5s model has an accuracy improvement of 3.1%, a recall rate improvement of 1%, and an average precision improvement of 2.5%, indicating the superiority of the improved YOLOv5s model in identifying transmission line hidden danger targets.
[0045] The method for detecting hidden danger targets in transmission lines provided in the embodiments of the present application automatically identifies transmission line images with hidden danger targets in a transmission line image sequence through a pre-trained deep learning model, without the need for manual detection, and is efficient and convenient. The deep learning model can utilize the self-attention mechanism to reconstruct the feature maps in the network, highlight important features, and improve the model's generalization ability for different scenarios. The FReLU nonlinear activation function is used to enhance the model's spatial sensitivity, so that ordinary convolution operations can also capture complex visual layouts, while giving the model pixel-level modeling capabilities, thereby greatly improving recognition accuracy and ensuring the safe operation of transmission lines.
[0046] Based on the above embodiments, the present application provides a device for detecting hidden danger targets in power transmission lines. The device 100 for detecting hidden danger targets in power transmission lines can be applied to Figures 1 to 6 In the corresponding embodiment of the power transmission line hidden danger target detection method. Please refer to Figure 7 The transmission line hidden danger target detection device 100 includes:
[0047] An acquisition module 101 is configured to acquire a sequence of transmission line images;
[0048] The recognition module 102 is configured to input a sequence of transmission line images into a pre-trained deep learning model to identify transmission line images with hidden danger targets. The deep learning model is a YOLOv5s model. A self-attention mechanism unit is set between the Backbone network and the Neck network in the YOLOv5s model. The YOLOv5s model uses an FReLU nonlinear activation function, which is obtained by expanding the ReLU function and the PReLU function into a two-dimensional activation function.
[0049] Optionally, in some embodiments of the present application, the self-attention mechanism unit includes a channel attention subunit and a spatial attention subunit. The channel attention subunit is configured to process the input feature map to obtain a channel attention map, and the spatial attention subunit is configured to obtain an output feature map based on the input feature map and the channel attention map.
[0050] Optionally, in some embodiments of the present application, the channel attention subunit is specifically used to input the input feature map into the maximum pooling channel and the average pooling channel respectively, and generate an intermediate vector; the intermediate vector is passed through a multi-layer perceptron to obtain a feature vector, and the feature vector is multiplied item by item and nonlinearly activated to obtain a channel attention map.
[0051] Optionally, in some embodiments of the present application, the spatial attention sub-unit is specifically used to process the multiplication result to obtain a spatial attention map, where the operation objects of the multiplication result are the input feature map and the channel attention map; and, the spatial attention map is multiplied with the multiplication result again to obtain an output feature map.
[0052] Optionally, in some embodiments of the present application, the spatial attention subunit is further specifically used to perform maximum pooling and average pooling operations on the multiplication results to generate an intermediate feature map; and perform convolution operations and nonlinear activation on the intermediate feature map to obtain a spatial attention map.
[0053] Optionally, in some embodiments of the present application, the activation threshold of the FReLU nonlinear activation function is a funnel feature map generated by performing a two-dimensional convolution operation on the input feature map.
[0054] Optionally, in some embodiments of the present application, the identification module 102 is specifically used to obtain transmission line hidden danger images of different voltage levels;
[0055] Label the types of transmission line hidden danger images and divide them into training and test sets;
[0056] The deep learning architecture is trained using the training set, and the training results are verified and weighted using the test set to obtain a deep learning model.
[0057] It should be noted that, for the description of the same steps and contents in this embodiment as those in other embodiments, reference can be made to the description in other embodiments and will not be repeated here.
[0058] The transmission line hidden danger target detection device provided in the embodiment of the present application automatically identifies transmission line images with hidden danger targets in a transmission line image sequence through a pre-trained deep learning model, without the need for manual detection, and is efficient and convenient. The deep learning model can use the self-attention mechanism to reconstruct the feature map in the network, highlight important features, and improve the model's generalization ability for different scenarios. It also uses the FReLU nonlinear activation function to enhance the model's spatial sensitivity, so that ordinary convolution operations can also capture complex visual layouts, while giving the model pixel-level modeling capabilities, thereby greatly improving recognition accuracy and ensuring the safe operation of transmission lines.
[0059] Based on the above embodiments, the present application provides an electronic device. Figure 8 The electronic device 200 may include a processor 201 and a memory 202. The memory 202 stores at least one instruction, at least one program, code set or instruction set, which is loaded and executed by the processor 201 to implement Figures 1 to 6 The steps of the transmission line hidden danger target detection method of the corresponding embodiment.
[0060] As another aspect, the present invention provides a computer-readable storage medium for storing program code for executing the aforementioned Figures 1 to 6 Any implementation manner of the transmission line hidden danger target detection method of the corresponding embodiment.
[0061] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.
[0062] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the modules is merely a logical function division. In actual implementation, there may be other division methods, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. On the other hand, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms. The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of this embodiment.
[0063] In addition, the functional modules in the various embodiments of the present application may be integrated into a single processing unit, or each module may exist physically separately, or two or more units may be integrated into a single module. The aforementioned integrated units may be implemented in the form of hardware or in the form of software functional units. If the integrated units are implemented in the form of software functional units and sold or used as independent products, they may be stored in a computer-readable storage medium.
[0064] Based on this understanding, the technical solution of this application, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the power transmission line hidden danger target detection method of each embodiment of this application. The aforementioned storage medium includes: a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc., various media that can store program code.
[0065] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, 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, they should be considered to be within the scope of this specification.
[0066] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.
Claims
1. A method for detecting hidden danger targets in power transmission lines, characterized in that: The transmission line hidden danger target detection method comprises: Acquire a transmission line image sequence; The transmission line image sequence is input into a pre-trained deep learning model to identify transmission line images with hidden danger targets. The deep learning model is a YOLOv5s model. A self-attention mechanism unit is arranged between the Backbone network and the Neck network in the YOLOv5s model. The YOLOv5s model adopts the FReLU nonlinear activation function, which is obtained by expanding the ReLU function and the PReLU function into a two-dimensional activation function.
2. The method for detecting hidden dangers in power transmission lines according to claim 1, characterized in that: The self-attention mechanism unit includes a channel attention subunit and a spatial attention subunit. The channel attention subunit is configured to process the input feature map to obtain a channel attention map, and the spatial attention subunit is configured to obtain an output feature map based on the input feature map and the channel attention map.
3. The method for detecting hidden dangers in power transmission lines according to claim 2, characterized in that: The channel attention subunit is specifically used to input the input feature map into the maximum pooling channel and the average pooling channel respectively, and generate an intermediate vector; obtain a feature vector by passing the intermediate vector through a multi-layer perceptron, and perform item-by-item multiplication and nonlinear activation on the feature vector to obtain the channel attention map.
4. The method for detecting hidden dangers in power transmission lines according to claim 2, characterized in that: The spatial attention subunit is specifically used to process the multiplication result to obtain a spatial attention map, where the operation objects of the multiplication result are the input feature map and the channel attention map; and, multiply the spatial attention map and the multiplication result again to obtain the output feature map.
5. The method for detecting hidden danger targets in power transmission lines according to claim 4, characterized in that: The spatial attention subunit is also specifically used to perform maximum pooling and average pooling operations on the multiplication results to generate an intermediate feature map; and perform convolution operations and nonlinear activation on the intermediate feature map to obtain the spatial attention map.
6. The method for detecting hidden danger targets in power transmission lines according to any one of claims 1 to 5, characterized in that: The activation threshold of the FReLU nonlinear activation function is a funnel feature map generated by performing a two-dimensional convolution operation on the input feature map.
7. The method for detecting hidden danger targets in power transmission lines according to claim 6, characterized in that: The deep learning model is trained through the following steps: Obtain hidden danger images of transmission lines at different voltage levels; Labeling the transmission line hidden danger images by type and dividing them into training sets and test sets; The deep learning architecture is trained using the training set, and the training results are verified and weighted using the test set to obtain the deep learning model.
8. A device for detecting hidden dangers in power transmission lines, characterized in that: The transmission line hidden danger target detection device comprises: an acquisition module configured to acquire a sequence of transmission line images; The recognition module is configured to input the transmission line image sequence into a pre-trained deep learning model to identify transmission line images with hidden danger targets, wherein the deep learning model is a YOLOv5s model, a self-attention mechanism unit is provided between the Backbone network and the Neck network in the YOLOv5s model, and the YOLOv5s model adopts the FReLU nonlinear activation function, which is obtained by expanding the ReLU function and the PReLU function into a two-dimensional activation function.
9. An electronic device, characterized in that: The electronic device includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, code set or instruction set, and the instruction, program, code set or instruction set is loaded and executed by the processor to implement the steps of the transmission line hidden danger target detection method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of the power transmission line hidden danger target detection method according to any one of claims 1 to 7.