A distribution network cable identification method and device based on semantic segmentation

Abstract

The present application relates to a distribution network cable identification method, device, computer equipment, storage medium and computer program product based on semantic segmentation. The method inputs an infrared image of the area where the distribution network cable to be identified is located into a recognition model. The recognition model is obtained by training the distribution network cable infrared image data set based on the semantic segmentation model. The recognition model includes a first convolution unit, a first deconvolution unit, a second convolution unit, a second deconvolution unit and an output unit from input to output. Based on the pixel-level classification result of the infrared image output by the recognition model, the image of the distribution network cable to be identified is identified from the infrared image. The method can realize the identification of the distribution network cable to be identified in a complex environment. The recognition model adopts a creatively designed stacked semantic segmentation model structure, and the recognition efficiency and accuracy are high.

Description

Technical Field

[0001] The present application relates to the field of cable technology, and in particular to a distribution network cable identification method, device, computer equipment, storage medium and computer program product based on semantic segmentation. Background Art

[0002] In recent years, with the continuous expansion of the scale of urban distribution networks and the continuous acceleration of overhead line undergrounding projects, the number of distribution cables has increased rapidly. Cables are used for power transmission and energy distribution, and their health and stability have become an important guarantee for the safe and reliable operation of urban power grids.

[0003] Distribution network cables (including cable bodies and their accessories) are often stored underground. The underground temperature environment is complex and the interference is large, which makes it extremely difficult to identify distribution network cables. The difficulty in identifying distribution network cables has led to manual identification by maintenance personnel, which requires high professionalism and work experience of maintenance personnel. The identification results are also subjective, which increases the difficulty of equipment operation and maintenance. Summary of the invention

[0004] Based on this, it is necessary to provide a distribution network cable identification method, device, computer equipment, storage medium and computer program product based on semantic segmentation to address the above technical problems.

[0005] In a first aspect, the present application provides a distribution network cable identification method based on semantic segmentation. The method comprises:

[0006] Acquire an infrared image of the area where the distribution network cable to be identified is located, the infrared image including the image of the distribution network cable to be identified and the background area;

[0007] The infrared image is input into the recognition model, where the recognition model is trained based on the semantic segmentation model using the infrared image data set of the distribution network cable. The recognition model includes a first convolution unit, a first deconvolution unit, a second convolution unit, a second deconvolution unit, and an output unit from input to output.

[0008] Based on the pixel-level classification result of the infrared image output by the recognition model, the image of the distribution network cable to be identified is obtained from the infrared image.

[0009] In one of the embodiments, the first convolution unit is used to perform convolution processing on the infrared image of the distribution network cable to be identified to obtain an underlying feature map; the first deconvolution unit is used to perform deconvolution processing on the underlying feature map output by the first convolution unit to obtain a global feature map; the second convolution unit is used to perform convolution processing on the global feature map output by the first deconvolution unit to obtain an image feature map; the second deconvolution unit is used to perform deconvolution processing on the image feature map output by the second convolution unit to obtain a semantic segmentation result map, and the image size of the semantic segmentation result map is the same as the image size of the infrared image of the distribution network cable to be identified; the output unit uses a fully connected neural network and a Softmax output layer to classify the semantic segmentation result map output by the second deconvolution unit to output a pixel-level classification result of the infrared image.

[0010] In one of the embodiments, the first convolution unit includes two convolution layers, one pooling layer, two convolution layers, one pooling layer, three convolution layers and one pooling layer from input to output; the first deconvolution unit includes one deconvolution layer, one padding layer, one convolution layer, one normalization layer, three convolution layers and one pooling layer from input to output; the second convolution unit includes two convolution sub-units from input to output, each of which includes three convolution layers and one pooling layer; the second deconvolution unit includes four deconvolution sub-units from input to output, each of which includes one deconvolution layer, one padding layer, one convolution layer and one normalization layer.

[0011] In one embodiment, the recognition model is trained based on the TensorFlow framework.

[0012] In one of the embodiments, obtaining an image of a distribution network cable to be identified from an infrared image includes: fusing a pixel-level classification result of the infrared image output by a recognition model with the infrared image to obtain an image of the distribution network cable to be identified.

[0013] In one of the embodiments, the method also includes: obtaining an infrared image dataset of a distribution network cable, the infrared image dataset including several sample infrared images of the distribution network cable, each pixel of each sample infrared image having a corresponding pixel-level label, the pixel-level label being a background label for indicating that the pixel belongs to a background area, or a foreground label for indicating that the pixel belongs to a distribution network cable; and using the infrared image dataset to perform model training based on a semantic segmentation model to obtain a recognition model.

[0014] In a second aspect, the present application also provides a distribution network cable identification device based on semantic segmentation. The device includes:

[0015] An image acquisition module is used to acquire an infrared image of the area where the distribution network cable to be identified is located, and the infrared image includes an image of the distribution network cable to be identified and a background area;

[0016] An image input module is used to input the infrared image into a recognition model. The recognition model is trained based on a semantic segmentation model using an infrared image data set of a distribution network cable. The recognition model includes a first convolution unit, a first deconvolution unit, a second convolution unit, a second deconvolution unit, and an output unit from input to output.

[0017] The recognition module is used to identify the image of the distribution network cable to be identified from the infrared image based on the pixel-level classification result of the infrared image output by the recognition model.

[0018] In a third aspect, the present application further provides a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the distribution network cable identification method based on semantic segmentation provided in the first aspect are implemented.

[0019] In a fourth aspect, the present application further provides a computer-readable storage medium, wherein a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the distribution network cable identification method based on semantic segmentation provided in the first aspect are implemented.

[0020] In a fifth aspect, the present application further provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the distribution network cable identification method based on semantic segmentation provided in the first aspect.

[0021] The above-mentioned distribution network cable identification method, device, computer equipment, storage medium and computer program product based on semantic segmentation use the recognition model obtained by training based on the semantic segmentation model to obtain pixel-level classification results from the infrared image of the distribution network cable to be identified, so as to identify the image of the distribution network cable to be identified, realize the identification of the distribution network cable to be identified in a complex environment, and facilitate the subsequent temperature feature extraction and research. The recognition model adopts a creatively designed cascaded semantic segmentation model structure, and has high recognition efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 The present invention is a flowchart of a method for identifying distribution network cables based on semantic segmentation according to an embodiment.

[0023] Figure 2 A schematic diagram of the model structure of two convolution units and two deconvolution units in an identification model in an embodiment.

[0024] Figure 3It is a schematic diagram of image fusion of an infrared image and a pixel-level classification result of the infrared image in one embodiment.

[0025] Figure 4 The present invention is a structural block diagram of a distribution network cable identification device based on semantic segmentation in one embodiment.

[0026] Figure 5 It is a structural block diagram of a distribution network cable identification device based on semantic segmentation in another embodiment.

[0027] Figure 6 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0029] In one embodiment, Figure 1 As shown, a distribution network cable identification method based on semantic segmentation is provided. This embodiment takes the method applied to a terminal as an example. It can be understood that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0030] Step 102: Acquire an infrared image of the area where the distribution network cable to be identified is located.

[0031] The distribution network cable in this application specifically includes cables and their accessories. The acquired infrared image includes images of the distribution network cable to be identified and the background area, and the acquired infrared image is an RGB color image. The distribution network cable to be identified is generally in a more complex environment, and the temperature difference between the distribution network cable to be identified and the surrounding background area is small, so it is difficult to directly identify the distribution network cable to be identified from the infrared image based on the temperature information.

[0032] Step 104, inputting the infrared image into the recognition model.

[0033] Among them, the recognition model includes a first convolution unit, a first deconvolution unit, a second convolution unit, a second deconvolution unit and an output unit from input to output. The recognition model is pre-trained based on a semantic segmentation model using an infrared image dataset of distribution network cables. After the infrared image is input into the recognition model, the output is a pixel-level classification result of the infrared image, that is, a result of classifying each pixel in the infrared image as belonging to the background area or the distribution network cable to be identified.

[0034] Step 106: Based on the pixel-level classification result of the infrared image output by the recognition model, an image of the distribution network cable to be identified is identified from the infrared image.

[0035] In the above-mentioned distribution network cable identification method based on semantic segmentation, the recognition model trained based on the semantic segmentation model is used to obtain pixel-level classification results from the infrared image of the distribution network cable to be identified, so that the image of the distribution network cable to be identified can be identified, and the identification of the distribution network cable to be identified in a complex environment is realized, which is convenient for subsequent temperature feature extraction and research. The recognition model adopts a creatively designed cascaded semantic segmentation model structure, and the recognition efficiency and accuracy are high.

[0036] In one embodiment, after the infrared image is input into the recognition model in the above step 104, the processing process of the infrared image by each unit in the recognition model is as follows: the first convolution unit is used to perform convolution processing on the infrared image of the distribution network cable to be identified to obtain the underlying feature map. The first deconvolution unit is used to perform deconvolution processing on the underlying feature map output by the first convolution unit to obtain the global feature map, that is, to restore the high-level features and global features from the underlying feature map. The second convolution unit is used to perform convolution processing on the global feature map output by the first deconvolution unit to obtain the image feature map, that is, to perform convolution operation again to extract features. The second deconvolution unit is used to perform deconvolution processing on the image feature map output by the second convolution unit to obtain a semantic segmentation result map, and the image size of the semantic segmentation result map is the same as the image size of the infrared image of the distribution network cable to be identified. The output unit uses a fully connected neural network and a Softmax output layer to classify the semantic segmentation result map output by the second deconvolution unit and output the pixel-level classification result of the infrared image.

[0037] In one embodiment, Figure 2 The structural diagram of each convolution unit and deconvolution unit in the recognition model shown. The first convolution unit includes two convolution layers, one pooling layer, two convolution layers, one pooling layer, three convolution layers and one pooling layer from input to output. The first deconvolution unit includes one deconvolution layer, one padding layer, one convolution layer, one normalization layer (Normalization), three convolution layers and one pooling layer from input to output. The second convolution unit includes two convolution sub-units from input to output, and each convolution sub-unit includes three convolution layers and one pooling layer. The second deconvolution unit includes four deconvolution sub-units from input to output, and each deconvolution sub-unit includes one deconvolution layer, one padding layer, one convolution layer and one normalization layer. Figure 2Taking the input infrared image Input with a size of 416*416*3 including three RGB channels as an example, the size becomes smaller after being processed by the first convolution unit, becomes larger after being processed by the first deconvolution unit, becomes smaller after being processed by the second convolution unit, and becomes larger after being processed by the second deconvolution unit, and finally outputs a semantic segmentation result map Output of 416*416*3 of the same size as the infrared image.

[0038] In the above step 106, the pixel-level classification result of the infrared image output by the recognition model is only the image segmentation result, so the pixel-level classification result of the infrared image output by the recognition model is fused with the infrared image to extract the image of the distribution network cable to be identified, such as Figure 3 In the final image of the distribution network cable to be identified, the distribution network cable to be identified has a large degree of distinction from the background area, which is convenient for temperature feature extraction.

[0039] In the above-mentioned distribution network cable identification method based on semantic segmentation, in one embodiment, before using the recognition model, it also includes a method for training to obtain the recognition model: obtaining an infrared image dataset of the distribution network cable, the infrared image dataset includes several sample infrared images of the distribution network cable, each pixel of each sample infrared image has a corresponding pixel-level label, and the pixel-level label is a background label used to indicate that the pixel belongs to the background area, or a foreground label used to indicate that the pixel belongs to the distribution network cable. The infrared image dataset is used to perform model training based on the semantic segmentation model to obtain a recognition model. The model structure of the semantic segmentation model used here is the same as the model structure of the recognition model introduced above, and this application will not repeat it. The infrared image dataset is input into the semantic segmentation model, and the pixel-level classification result output by the semantic segmentation model is compared with the pixel-level label of each pixel of the sample infrared image and the model parameters of the semantic segmentation model are updated to complete the model training.

[0040] Optionally, taking full account of the characteristics of the distribution network cables, the recognition model is trained based on the TensorFlow framework.

[0041] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0042] Based on the same inventive concept, the embodiment of the present application also provides a distribution network cable identification device based on semantic segmentation for implementing the distribution network cable identification method based on semantic segmentation involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more distribution network cable identification device embodiments based on semantic segmentation provided below can refer to the limitations of the distribution network cable identification method based on semantic segmentation above, and will not be repeated here.

[0043] In one embodiment, Figure 4 As shown, a distribution network cable identification device based on semantic segmentation is provided, comprising: an image acquisition module, an image input module and a recognition module, wherein:

[0044] The image acquisition module 410 is used to acquire an infrared image of the area where the distribution network cable to be identified is located. The infrared image includes images of the distribution network cable to be identified and the background area.

[0045] The image input module 420 is used to input the infrared image into the recognition model. The recognition model is trained based on the semantic segmentation model using the infrared image data set of the distribution network cable. The recognition model includes a first convolution unit, a first deconvolution unit, a second convolution unit, a second deconvolution unit and an output unit from input to output.

[0046] The recognition module 430 is used to recognize the image of the distribution network cable to be recognized from the infrared image based on the pixel-level classification result of the infrared image output by the recognition model.

[0047] In one embodiment, the first convolution unit is used to perform convolution processing on the infrared image of the distribution network cable to be identified to obtain an underlying feature map; the first deconvolution unit is used to perform deconvolution processing on the underlying feature map output by the first convolution unit to obtain a global feature map; the second convolution unit is used to perform convolution processing on the global feature map output by the first deconvolution unit to obtain an image feature map; the second deconvolution unit is used to perform deconvolution processing on the image feature map output by the second convolution unit to obtain a semantic segmentation result map, and the image size of the semantic segmentation result map is the same as the image size of the infrared image of the distribution network cable to be identified; the output unit uses a fully connected neural network and a Softmax output layer to classify the semantic segmentation result map output by the second deconvolution unit to output a pixel-level classification result of the infrared image.

[0048] In one embodiment, the first convolution unit includes two convolution layers, one pooling layer, two convolution layers, one pooling layer, three convolution layers and one pooling layer from input to output; the first deconvolution unit includes one deconvolution layer, one padding layer, one convolution layer, one normalization layer, three convolution layers and one pooling layer from input to output; the second convolution unit includes two convolution sub-units from input to output, each of which includes three convolution layers and one pooling layer; the second deconvolution unit includes four deconvolution sub-units from input to output, each of which includes one deconvolution layer, one padding layer, one convolution layer and one normalization layer.

[0049] In one embodiment, the recognition model is trained based on the TensorFlow framework.

[0050] In one embodiment, the recognition module is further used to perform image fusion on the infrared image by combining the pixel-level classification result of the infrared image output by the recognition model with the infrared image to obtain an image of the distribution network cable to be identified.

[0051] In one embodiment, please refer to Figure 5 The device also includes a model training module 440, which is used to obtain an infrared image dataset of a distribution network cable, the infrared image dataset including several sample infrared images of the distribution network cable, each pixel of each sample infrared image having a corresponding pixel-level label, the pixel-level label being a background label for indicating that the pixel belongs to a background area, or a foreground label for indicating that the pixel belongs to a distribution network cable; a recognition model is obtained by performing model training based on a semantic segmentation model using the infrared image dataset.

[0052] Each module in the above-mentioned distribution network cable identification device based on semantic segmentation can be implemented in whole or in part by software, hardware and their combination. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0053] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 6 As shown. The computer device includes a processor, a memory and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store pre-trained recognition models. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a distribution network cable identification method based on semantic segmentation is implemented.

[0054] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0055] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0056] Acquire an infrared image of the area where the distribution network cable to be identified is located, the infrared image including the image of the distribution network cable to be identified and the background area;

[0057] The infrared image is input into the recognition model, where the recognition model is trained based on the semantic segmentation model using the infrared image data set of the distribution network cable. The recognition model includes a first convolution unit, a first deconvolution unit, a second convolution unit, a second deconvolution unit, and an output unit from input to output.

[0058] Based on the pixel-level classification result of the infrared image output by the recognition model, the image of the distribution network cable to be identified is obtained from the infrared image.

[0059] In one embodiment, when the processor executes the computer program, the following steps are also implemented: the pixel-level classification result of the infrared image output by the recognition model is image-fused with the infrared image to obtain an image of the distribution network cable to be identified.

[0060] In one embodiment, when the processor executes the computer program, the following steps are also implemented: obtaining an infrared image dataset of a distribution network cable, the infrared image dataset including several sample infrared images of the distribution network cable, each pixel of each sample infrared image having a corresponding pixel-level label, the pixel-level label being a background label for indicating that the pixel belongs to a background area, or a foreground label for indicating that the pixel belongs to a distribution network cable; and using the infrared image dataset to perform model training based on a semantic segmentation model to obtain a recognition model.

[0061] 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 following steps are implemented:

[0062] Acquire an infrared image of the area where the distribution network cable to be identified is located, the infrared image including the image of the distribution network cable to be identified and the background area;

[0063] The infrared image is input into the recognition model, where the recognition model is trained based on the semantic segmentation model using the infrared image data set of the distribution network cable. The recognition model includes a first convolution unit, a first deconvolution unit, a second convolution unit, a second deconvolution unit, and an output unit from input to output.

[0064] Based on the pixel-level classification result of the infrared image output by the recognition model, the image of the distribution network cable to be identified is obtained from the infrared image.

[0065] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: the pixel-level classification result of the infrared image output by the recognition model is image-fused with the infrared image to obtain an image of the distribution network cable to be identified.

[0066] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtaining an infrared image dataset of a distribution network cable, the infrared image dataset including several sample infrared images of the distribution network cable, each pixel of each sample infrared image having a corresponding pixel-level label, the pixel-level label being a background label for indicating that the pixel belongs to a background area, or a foreground label for indicating that the pixel belongs to a distribution network cable; and using the infrared image dataset to perform model training based on a semantic segmentation model to obtain a recognition model.

[0067] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0068] Acquire an infrared image of the area where the distribution network cable to be identified is located, the infrared image including the image of the distribution network cable to be identified and the background area;

[0069] The infrared image is input into the recognition model, where the recognition model is trained based on the semantic segmentation model using the infrared image data set of the distribution network cable. The recognition model includes a first convolution unit, a first deconvolution unit, a second convolution unit, a second deconvolution unit, and an output unit from input to output.

[0070] Based on the pixel-level classification result of the infrared image output by the recognition model, the image of the distribution network cable to be identified is obtained from the infrared image.

[0071] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: the pixel-level classification result of the infrared image output by the recognition model is image-fused with the infrared image to obtain an image of the distribution network cable to be identified.

[0072] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtaining an infrared image dataset of a distribution network cable, the infrared image dataset including several sample infrared images of the distribution network cable, each pixel of each sample infrared image having a corresponding pixel-level label, the pixel-level label being a background label for indicating that the pixel belongs to a background area, or a foreground label for indicating that the pixel belongs to a distribution network cable; and using the infrared image dataset to perform model training based on a semantic segmentation model to obtain a recognition model.

[0073] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0074] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0075] The technical features of the above embodiments may be arbitrarily combined. 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.

[0076] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made 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 shall be subject to the attached claims.

Claims

1. A distribution network cable identification method based on semantic segmentation, characterized in that: The method comprises: Acquire an infrared image of the area where the distribution network cable to be identified is located, the infrared image including the image of the distribution network cable to be identified and the background area; Inputting the infrared image into a recognition model, wherein the recognition model is trained based on a semantic segmentation model using an infrared image data set of a distribution network cable, and the recognition model includes, from input to output, a first convolution unit, a first deconvolution unit, a second convolution unit, a second deconvolution unit, and an output unit; Based on the pixel-level classification result of the infrared image output by the recognition model, identifying and obtaining the image of the distribution network cable to be identified from the infrared image; The first convolution unit includes two convolution layers, one pooling layer, two convolution layers, one pooling layer, three convolution layers and one pooling layer from input to output; The first deconvolution unit includes, from input to output, a deconvolution layer, a padding layer, a convolution layer, a normalization layer, three convolution layers and a pooling layer; The second convolution unit includes two convolution subunits from input to output, and each convolution subunit includes three convolution layers and one pooling layer; The second deconvolution unit includes four deconvolution subunits from input to output, and each deconvolution subunit includes a deconvolution layer, a padding layer, a convolution layer and a normalization layer.

2. The method according to claim 1, characterized in that The first convolution unit is used to perform convolution processing on the infrared image of the distribution network cable to be identified to obtain an underlying feature map; The first deconvolution unit is used to perform deconvolution processing on the bottom feature map output by the first convolution unit to obtain a global feature map; The second convolution unit is used to perform convolution processing on the global feature map output by the first deconvolution unit to obtain an image feature map; The second deconvolution unit is used to perform deconvolution processing on the image feature map output by the second convolution unit to obtain a semantic segmentation result map, and the image size of the semantic segmentation result map is the same as the image size of the infrared image of the distribution network cable to be identified; The output unit uses a fully connected neural network and a Softmax output layer to classify the semantic segmentation result map output by the second deconvolution unit and output a pixel-level classification result of the infrared image.

3. The method according to claim 1, characterized in that The infrared image is an RGB color image.

4. The method according to claim 1, characterized in that: The recognition model is trained based on the TensorFlow framework.

5. The method according to any one of claims 1 to 4, characterized in that: The step of identifying the image of the distribution network cable to be identified from the infrared image comprises: The pixel-level classification result of the infrared image output by the recognition model is fused with the infrared image to obtain the image of the distribution network cable to be identified.

6. The method according to any one of claims 1 to 4, characterized in that: The method further comprises: Acquire an infrared image dataset of a distribution network cable, wherein the infrared image dataset includes a plurality of sample infrared images of the distribution network cable, each pixel of each sample infrared image has a corresponding pixel-level label, and the pixel-level label is a background label used to indicate that the pixel belongs to a background area, or a foreground label used to indicate that the pixel belongs to a distribution network cable; The recognition model is obtained by performing model training based on a semantic segmentation model using the infrared image data set.

7. A distribution network cable identification device based on semantic segmentation, characterized in that: The device comprises: An image acquisition module, used to acquire an infrared image of the area where the distribution network cable to be identified is located, wherein the infrared image includes an image of the distribution network cable to be identified and an image of a background area; An image input module, used for inputting the infrared image into a recognition model, wherein the recognition model is trained based on a semantic segmentation model using an infrared image data set of a distribution network cable, and the recognition model includes, from input to output, a first convolution unit, a first deconvolution unit, a second convolution unit, a second deconvolution unit, and an output unit; A recognition module, configured to recognize and obtain the image of the distribution network cable to be identified from the infrared image based on the pixel-level classification result of the infrared image output by the recognition model; The first convolution unit includes two convolution layers, one pooling layer, two convolution layers, one pooling layer, three convolution layers and one pooling layer from input to output; The first deconvolution unit includes, from input to output, a deconvolution layer, a padding layer, a convolution layer, a normalization layer, three convolution layers and a pooling layer; The second convolution unit includes two convolution subunits from input to output, and each convolution subunit includes three convolution layers and one pooling layer; The second deconvolution unit includes four deconvolution subunits from input to output, and each deconvolution subunit includes a deconvolution layer, a padding layer, a convolution layer and a normalization layer.

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

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

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

Images