Infrared Image Defect Identification Method for Power Switchgear Equipment

Abstract

This invention relates to the field of image recognition technology, and more particularly to a method for identifying defects in infrared images of power switchgear equipment. The method includes the following steps: acquiring an infrared image of the power switchgear equipment to be inspected; constructing an infrared image defect detection model based on a YOLOv10 detection model, comprising a SPAN attention module and a c2f feature extraction module; and performing defect detection on the infrared image of the power switchgear equipment based on the infrared image defect detection model to obtain the infrared image defect detection result. This invention improves upon the existing YOLOv10 detection model, effectively enhancing the feature extraction capability and detection accuracy of the infrared image defect detection model. Simultaneously, it does not significantly increase the computational burden compared to existing structures, ensuring the running speed of the detection model on infrared detection hardware, and effectively solving the problem of insufficient accuracy in detecting infrared image defects in power switchgear equipment using existing methods.

Description

Technical Field

[0001] This invention relates to the field of image recognition technology, and in particular to a method for identifying defects in infrared images of power switchgear equipment. Background Technology

[0002] With the continuous development of infrared imaging technology, infrared images are being used more and more widely in various fields. In the power environment, infrared images are often used to analyze the operating status of power equipment. For power switchgear, which is installed at high locations and in complex environments, using long-distance infrared imaging for defect detection is currently a common detection method.

[0003] However, due to the inherent characteristics of infrared images, such as low resolution, high noise, and low contrast, traditional image recognition algorithms, such as Support Vector Machine (SVM) and K-Nearest Neighbors (KNN), are difficult to achieve ideal results. These traditional methods mainly rely on manually extracted features, which perform poorly when processing high-noise, low-contrast infrared images and are difficult to achieve real-time detection.

[0004] In recent years, the YOLO series of algorithms, with their end-to-end, real-time, and efficient target detection capabilities, have been applied to infrared image detection. Through end-to-end feature learning, these methods have improved detection performance to a certain extent and achieved great success in the visible light image field. The latest YOLOv10 algorithm has further optimized the model's detection accuracy and speed. However, when applied to the low-contrast characteristics of infrared images, the YOLOv10 algorithm is not sensitive enough to the low-contrast target features unique to infrared images, and cannot fully utilize the spectral information of infrared images for target detection, resulting in a high false negative rate. Secondly, although the YOLOv10 algorithm improves detection accuracy, its computational complexity is high and the number of model parameters is too large, making it difficult to apply in real time on resource-constrained infrared imaging equipment for power switchgear detection.

[0005] Therefore, it is necessary to study and optimize the problem of infrared image defect identification in power switchgear equipment. Summary of the Invention

[0006] The present invention aims to solve the technical problem that existing image detection methods are insufficient in terms of the accuracy of infrared image defect detection for power switchgear.

[0007] To address the aforementioned technical problems, this invention provides a method for infrared image defect identification in power switchgear equipment, comprising the following steps:

[0008] Acquire infrared images of the power switchgear equipment to be inspected;

[0009] Based on the YOLOv10 detection model, an infrared image defect detection model including a SPAN attention module and a c2f feature extraction module was constructed.

[0010] Based on the infrared image defect detection model, defects are detected in the infrared images of the power switchgear equipment to obtain infrared image defect detection results.

[0011] Furthermore, the infrared image defect detection model includes an input network, a backbone network, a neck network, and a head network, wherein:

[0012] The input network is used to preprocess the image data fed into the infrared image defect detection model;

[0013] The backbone network is used to extract the first image feature from the image data;

[0014] The neck network is used to process and fuse the first image features to obtain the second image features;

[0015] The head network is used to perform target detection based on the second image features and output the target detection result of a specific target in the image data.

[0016] Furthermore, the SPAN attention module is located in the neck network, and the neck network includes, in sequence according to the input-output logic order of the infrared image defect detection model:

[0017] The path aggregation layer includes multiple processing units consisting of an upsampling module, a feature splicing module, and a first C2F module connected in sequence.

[0018] The feature fusion layer includes, in sequence, a convolution module, a first feature concatenation module, a second C2F module, a downsampling module, a second feature concatenation module, and a C2FCIB feature fusion module;

[0019] The SPAN attention module comprises three modules, which are respectively positioned between the path aggregation layer and the feature fusion layer, after the second c2f module, and after the c2fcib feature fusion module;

[0020] The neck network includes three outputs, with the first output located at the output of the path aggregation layer, used to output large-size image features extracted by the path aggregation layer.

[0021] The second output terminal is located at the output terminal of the second c2f module and is used to output the medium-sized image features extracted by the second c2f module;

[0022] The third output is located after the feature fusion layer and is used to output the small-sized image features extracted by the feature fusion layer.

[0023] The large-size image features, the medium-size image features, and the small-size image features are collectively input into the head network as the second image features output by the neck network.

[0024] Furthermore, the SPAN attention module performs pooling operations on the height and width of the input image features tensor to aggregate them, and obtains the first feature enhancement signals in the height and width directions through 1*1 convolution and regularization processing.

[0025] The SPAN attention module also includes a multi-scale convolutional unit composed of multiple convolutional kernels of different sizes. After the first feature enhancement signal is aggregated with the input image features, multi-scale feature extraction is performed through the multi-scale convolutional unit to obtain the second feature enhancement signal.

[0026] The SPAN attention module further includes a soft attention mechanism unit for extracting similarity weights between features in the input image features. The second feature enhancement signal and the similarity weights together serve as the output of the SPAN attention module.

[0027] Furthermore, the c2f feature extraction module replaces the c2f module in the backbone network of the original YOLOv10 detection model and forms the backbone network of the infrared image defect detection model.

[0028] Furthermore, the C2F feature extraction module processes the input image features through sequentially connected local average pooling and global average pooling to obtain local and global features. Then, after 1*1 convolution and adaptive average pooling, global attention weights are obtained.

[0029] The c2f feature extraction module further processes the output of local average pooling through a 1*1 convolution, and then activates it through a sigmoid activation function to obtain local attention weights.

[0030] The C2F feature extraction module further includes a soft attention mechanism unit, which is used to extract similarity weights between features in the input image features. The global attention weights and the local attention weights are combined and used together with the similarity weights as the output of the C2F feature extraction module.

[0031] Furthermore, when the global attention weight and the local attention weight are combined, a preset local weight parameter is used to control the combination ratio of the global attention weight and the local attention weight.

[0032] The beneficial effects achieved by this invention lie in proposing a method for infrared image defect identification of power switchgear equipment based on an improved YOLOv10 detection model. This method improves the existing YOLOv10 detection model by adding a SPAN attention module and a c2f feature extraction module to the network structure. These two modules effectively enhance the feature extraction capability and detection accuracy of the infrared image defect detection model, while not significantly increasing the computational burden compared to the existing structure, ensuring the running speed of the detection model on infrared detection hardware, and effectively solving the problem of insufficient accuracy in infrared image defect detection of power switchgear in existing methods. Attached Figure Description

[0033] Figure 1 This is a flowchart illustrating the steps of the infrared image defect identification method for power switchgear equipment provided in this embodiment of the invention.

[0034] Figure 2 This is a schematic diagram of the original YOLOv10 detection model provided in the embodiments of the present invention;

[0035] Figure 3 This is a schematic diagram of the neck network structure of the infrared image defect detection model provided in an embodiment of the present invention;

[0036] Figure 4 This is a schematic diagram of the SPAN attention module of the infrared image defect detection model provided in this embodiment of the invention;

[0037] Figure 5 This is a schematic diagram of the backbone network structure of the infrared image defect detection model provided in this embodiment of the invention;

[0038] Figure 6 This is a schematic diagram of the c2f feature extraction module structure of the infrared image defect detection model provided in this embodiment of the invention;

[0039] Figure 7 This is a schematic diagram showing the results of testing and comparing the infrared image defect detection model and related detection methods provided in this embodiment of the invention on different datasets. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of this invention clearer, the invention 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 merely illustrative and not intended to limit the invention.

[0041] For details, please refer to Figure 1 , Figure 1This is a schematic flowchart of the infrared image defect identification method for power switchgear equipment provided in an embodiment of the present invention. The infrared image defect identification method for power switchgear equipment includes the following steps:

[0042] S101. Obtain infrared images of the power switchgear equipment to be inspected;

[0043] S102. Based on the YOLOv10 detection model, construct an infrared image defect detection model that includes a SPAN attention module and a c2f feature extraction module.

[0044] S103. Based on the infrared image defect detection model, perform defect detection on the infrared image of the power switchgear equipment to obtain the infrared image defect detection result.

[0045] In step S101, the infrared image of the power switchgear equipment can be obtained by capturing images of the power switchgear equipment using a specific infrared image acquisition device.

[0046] In this embodiment of the invention, the infrared image defect detection model is obtained by structurally improving the original YOLOv10 detection model. The model structure of the original YOLOv10 detection model is as follows: Figure 2 As shown, the infrared image defect detection model in this embodiment of the invention is consistent with the original YOLOv10 detection model in terms of model construction. Specifically, the infrared image defect detection model includes an input network, a backbone network, a neck network, and a head network, wherein:

[0047] The input network is used to preprocess the image data fed into the infrared image defect detection model;

[0048] The backbone network is used to extract the first image feature from the image data;

[0049] The neck network is used to process and fuse the first image features to obtain the second image features;

[0050] The head network is used to perform target detection based on the second image features and output the target detection result of a specific target in the image data.

[0051] Compared to the original YOLOv10 detection model, this embodiment of the invention adds a SPAN (Spectral Precision Augmented Attention Network) and a C2f (C2f Precision Augmented Module) feature extraction module. Please refer to... Figure 3 , Figure 3This is a schematic diagram of the neck network structure of the infrared image defect detection model provided in this embodiment of the invention. The SPAN attention module is disposed in the neck network, and the neck network includes, in sequence according to the input-output logic order of the infrared image defect detection model:

[0052] The path aggregation layer includes multiple processing units consisting of an upsampling module, a feature concatenation module, and a first c2f module connected in sequence.

[0053] The feature fusion layer comprises, in sequence, a convolutional module (Conv), a first feature concatenation module, a second c2f module, a downsampling module, a second feature concatenation module, and a c2fcib feature fusion module;

[0054] The SPAN attention module comprises three modules, which are respectively positioned between the path aggregation layer and the feature fusion layer, after the second c2f module, and after the c2fcib feature fusion module;

[0055] The neck network includes three outputs, with the first output located at the output of the path aggregation layer, used to output large-size image features extracted by the path aggregation layer.

[0056] The second output terminal is located at the output terminal of the second c2f module and is used to output the medium-sized image features extracted by the second c2f module;

[0057] The third output is located after the feature fusion layer and is used to output the small-sized image features extracted by the feature fusion layer.

[0058] The large-size image features, the medium-size image features, and the small-size image features are collectively input into the head network as the second image features output by the neck network.

[0059] For details, please refer to Figure 4 , Figure 4 This is a schematic diagram of the SPAN attention module of the infrared image defect detection model provided in this embodiment of the invention. The SPAN attention module performs pooling operations on the height and width of the input image feature tensor to aggregate them, and obtains the first feature enhancement signals in the height and width directions through 1*1 convolution and regularization processing (GroupNorm). This global information extraction method helps the SPAN attention module to better capture the overall structure and global texture of the image, thereby improving the model's ability to capture long-distance dependencies. When facing complex visual tasks, this enhancement of the global field of view makes the model more expressive when processing large scenes or large objects.

[0060] The SPAN attention module further includes a multi-scale convolutional unit composed of multiple convolutional kernels of different sizes. After the first feature enhancement signal is aggregated with the input image features, multi-scale feature extraction is performed through the multi-scale convolutional unit to obtain the second feature enhancement signal. In this embodiment, the multi-scale convolutional unit is set to include convolutions of 1*7, 7*1, 1*11, and 11*1 to process features, extracting information from multiple receptive fields. Processing the input image with convolutional kernels of the same scale can effectively capture multi-scale feature information in the image. This multi-scale convolution operation is particularly suitable for processing image features containing different levels of detail, and can better extract small local features and large-scale global features. Compared to a traditional single convolutional kernel, the SPAN attention module in this embodiment can capture information from different receptive fields, enabling the model to more comprehensively understand the object structure and texture information in the image, thereby improving the accuracy of feature extraction.

[0061] The SPAN attention module also includes a soft attention mechanism unit (Softmax) for extracting similarity weights between features in the input image. The second feature enhancement signal and the similarity weights are used together as the output of the SPAN attention module. The soft attention mechanism is implemented through qkv(query, key, value) operations. This mechanism can dynamically adjust the importance of each image region. By calculating the similarity weights between features, attention is focused on more significant feature regions, improving the accuracy of feature selection. With the above structure, the SPAN attention module in this embodiment can more accurately extract key features of target objects in complex scenes and weaken the interference of background or irrelevant regions. This makes the module exhibit strong robustness in tasks such as object detection and image classification, and can improve the model's accuracy in recognizing target objects.

[0062] Please refer to Figure 5 , Figure 5 This is a schematic diagram of the backbone network structure of the infrared image defect detection model provided in this embodiment of the invention. The c2f feature extraction module replaces the c2f module in the backbone network of the original YOLOv10 detection model and forms the backbone network of the infrared image defect detection model.

[0063] For details, please refer to Figure 6 , Figure 6This is a schematic diagram of the C2F feature extraction module structure of the infrared image defect detection model provided in this embodiment of the invention. The C2F feature extraction module processes the input image features through sequentially connected local arv pooling and global arv pooling to obtain local features and global features. Then, after 1*1 convolution and adaptive mean pooling, global attention weights are obtained.

[0064] The C2F feature extraction module further processes the output of local average pooling through a 1*1 convolution, followed by activation using a sigmoid activation function to obtain local attention weights. The global features input to the C2F feature extraction module undergo local and global pooling, followed by further processing by global convolution. Local features, after local pooling, are directly processed by local convolution. Local convolution focuses on capturing local details of the image, while global convolution excels at processing the overall image structure and capturing full-text information. The combination of local and global convolutions can effectively adjust the number of channels in the feature map, perform multi-scale fusion, and allow for adaptive convolution kernel adjustment—dynamically selecting the kernel size based on the input size or feature type. This enables the model to flexibly adjust its receptive field to adapt to different input features. Furthermore, the combination of local and global feature attention integrates detailed information and global background, providing richer and more comprehensive features for the attention mechanism.

[0065] The C2F feature extraction module also includes a softmax attention mechanism unit, used to extract similarity weights between features in the input image. The global attention weights and local attention weights, combined with the similarity weights, serve as the output of the C2F feature extraction module. The softmax attention mechanism unit directly receives a branch from the input feature (X) for processing, performing self-attention calculations on the features. Through matrix multiplication between q, k, and v, the C2F feature extraction module can adaptively select important feature information. This approach helps improve the network's response to important regions, reduces attention to redundant information, and thus improves the accuracy of image feature extraction.

[0066] Furthermore, when the global attention weights and the local attention weights are combined, a preset local weight parameter controls the combination ratio of the global attention weights and the local attention weights. During implementation, this parameter can be manually adjusted to flexibly adapt to different image feature extraction tasks. For example, in tasks requiring emphasis on local details, increasing the local weights is more conducive to extracting finer features, while when global information is needed, increasing the global weights can make the network pay more attention to overall information and optimize feature representation.

[0067] The SPAN attention module proposed in this embodiment of the invention extracts global information through adaptive pooling, captures details and global features through multi-scale convolution, accurately focuses on important regions using a soft attention mechanism, and employs learnable weighted fusion of multi-scale features. Compared with traditional single convolution or fixed attention mechanisms, the SPAN attention module can flexibly handle complex scenes, more comprehensively understand image content, and dynamically optimize feature representation, significantly improving the model's accuracy and robustness.

[0068] The C2F feature extraction module, through the combination of local and global feature extraction and the integration of convolutional attention and soft attention, possesses powerful image feature extraction capabilities. This module design not only improves the model's comprehensive understanding of local and global information but also allows it to flexibly adapt to different image features and contextual information, demonstrating significant application prospects in computer vision tasks.

[0069] The results of testing and comparing the infrared image defect detection model and related detection methods proposed in this invention on different datasets are as follows: Figure 7 As shown, the detection model proposed in this embodiment of the invention ( Figure 7 The YOLOv10-DBB-c2f-PAM-SPAN method has higher detection accuracy than other existing detection methods and also has better detection performance when applied to the infrared image defect detection of power switchgear.

[0070] The beneficial effects achieved by this invention lie in proposing a method for infrared image defect identification of power switchgear equipment based on an improved YOLOv10 detection model. This method improves the existing YOLOv10 detection model by adding a SPAN attention module and a c2f feature extraction module to the network structure. These two modules effectively enhance the feature extraction capability and detection accuracy of the infrared image defect detection model, while not significantly increasing the computational burden compared to the existing structure, ensuring the running speed of the detection model on infrared detection hardware, and effectively solving the problem of insufficient accuracy in infrared image defect detection of power switchgear in existing methods.

[0071] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0072] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0073] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better embodiment. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present invention.

[0074] The embodiments of the present invention have been described above with reference to the accompanying drawings. The disclosed embodiments are merely preferred embodiments of the present invention. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many equivalent changes in form without departing from the spirit and scope of the claims of the present invention, and all such changes are within the protection scope of the present invention.

Claims

1. A method for infrared image defect identification of power switchgear equipment, characterized in that, Includes the following steps: Acquire infrared images of the power switchgear equipment to be inspected; Based on the YOLOv10 detection model, an infrared image defect detection model including a SPAN attention module and a c2f feature extraction module was constructed. Based on the infrared image defect detection model, defects are detected in the infrared image of the power switchgear equipment to obtain the infrared image defect detection results. The infrared image defect detection model includes an input network, a backbone network, a neck network, and a head network, wherein: The input network is used to preprocess the image data fed into the infrared image defect detection model; The backbone network is used to extract the first image feature from the image data; The neck network is used to process and fuse the first image features to obtain the second image features; The head network is used to perform target detection based on the second image features and output the target detection result of a specific target in the image data; The SPAN attention module is located in the neck network, which, according to the input-output logic order of the infrared image defect detection model, includes the following components in sequence: The path aggregation layer includes multiple processing units consisting of an upsampling module, a feature splicing module, and a first C2F module connected in sequence. The feature fusion layer includes, in sequence, a convolution module, a first feature concatenation module, a second C2F module, a downsampling module, a second feature concatenation module, and a C2FCIB feature fusion module; The SPAN attention module comprises three modules, which are respectively positioned between the path aggregation layer and the feature fusion layer, after the second c2f module, and after the c2fcib feature fusion module; The neck network includes three outputs, with the first output located at the output of the path aggregation layer, used to output large-size image features extracted by the path aggregation layer. The second output terminal is located at the output terminal of the second c2f module and is used to output the medium-sized image features extracted by the second c2f module; The third output is located after the feature fusion layer and is used to output the small-sized image features extracted by the feature fusion layer. The large-size image features, the medium-size image features, and the small-size image features are collectively input into the head network as the second image features output by the neck network. The SPAN attention module performs pooling operations on the height and width of the input image features tensor to aggregate them, and obtains the first feature enhancement signals in the height and width directions through 1*1 convolution and regularization. The SPAN attention module also includes a multi-scale convolutional unit composed of multiple convolutional kernels of different sizes. After the first feature enhancement signal is aggregated with the input image features, multi-scale feature extraction is performed through the multi-scale convolutional unit to obtain the second feature enhancement signal. The SPAN attention module further includes a soft attention mechanism unit for extracting similarity weights between features in the input image features. The second feature enhancement signal and the similarity weights together serve as the output of the SPAN attention module.

2. The infrared image defect identification method for power switchgear equipment according to claim 1, characterized in that, The c2f feature extraction module replaces the c2f module in the backbone network of the original YOLOv10 detection model and forms the backbone network of the infrared image defect detection model.

3. The infrared image defect identification method for power switchgear equipment according to claim 2, characterized in that, The C2F feature extraction module processes the input image features through sequential local average pooling and global average pooling to obtain local and global features. Then, after 1*1 convolution and adaptive average pooling, global attention weights are obtained. The c2f feature extraction module further processes the output of local average pooling through a 1*1 convolution, and then activates it through a sigmoid activation function to obtain local attention weights. The C2F feature extraction module further includes a soft attention mechanism unit, which is used to extract similarity weights between features in the input image features. The global attention weights and the local attention weights are combined and used together with the similarity weights as the output of the C2F feature extraction module.

4. The infrared image defect identification method for power switchgear equipment according to claim 3, characterized in that, When the global attention weight and the local attention weight are combined, a preset local weight parameter is used to control the combination ratio of the global attention weight and the local attention weight.

Images