CN114782839A

CN114782839A - Method, device and equipment for detecting power transmission line inspection fault and storage medium

Info

Publication number: CN114782839A
Application number: CN202210333321.8A
Authority: CN
Inventors: 杨喆; 何勇; 原瀚杰; 张雨; 董丽梦; 梁健波; 罗建斌; 谭麒; 姚健安; 邓浩光; 姜南; 陆勇生; 陆林
Original assignee: Guangdong Power Grid Co Ltd; Zhaoqing Power Supply Bureau of Guangdong Power Grid Co Ltd
Current assignee: Guangdong Power Grid Co Ltd; Zhaoqing Power Supply Bureau of Guangdong Power Grid Co Ltd
Priority date: 2022-03-31
Filing date: 2022-03-31
Publication date: 2022-07-22

Abstract

The application discloses a method, a device, equipment and a storage medium for detecting a power transmission line inspection fault, which are used for solving the technical problem that the real-time fault detection is difficult to realize in the existing fault detection method. The method comprises the following steps: acquiring a patrol image shot when the power transmission line is patrolled; extracting the features of the inspection image to obtain an output feature map; carrying out target detection on the output characteristic diagram to obtain a target detection result, wherein the target detection result comprises: category information and location information; decoding the position information to obtain a target boundary box; and outputting a fault detection result corresponding to the inspection image according to the target boundary frame and the category information.

Description

Method, device and equipment for detecting inspection faults of power transmission line and storage medium

Technical Field

The application relates to the technical field of power inspection, in particular to a method, a device, equipment and a storage medium for detecting inspection faults of a power transmission line.

Background

In an electric power system, daily inspection is a necessary means for ensuring safe and stable operation of a power grid. With the development of unmanned aerial vehicle technology, the inspection mode of mainly unmanned aerial vehicle inspection and auxiliary manual inspection becomes the mainstream inspection mode of inspection of transmission lines in China. Meanwhile, the unmanned aerial vehicle-based inspection image is used for detecting the faults of the power transmission line, and the current hot research direction is also formed.

The detection method based on deep learning is widely applied to fault detection of power routing inspection at present, and although the method has high detection performance, real-time fault detection is difficult to realize.

Disclosure of Invention

The application provides a method, a device, equipment and a storage medium for detecting a power transmission line inspection fault, which can realize real-time fault detection and solve the technical problem that the real-time fault detection is difficult to realize by the existing fault detection method.

The application provides a detection method for power transmission line inspection faults, which comprises the following steps:

acquiring a patrol image shot when the power transmission line is patrolled;

extracting the features of the inspection image to obtain an output feature map;

carrying out target detection on the output characteristic diagram to obtain a target detection result, wherein the target detection result comprises: category information and location information;

decoding the position information to obtain a target boundary box;

and outputting a fault detection result corresponding to the inspection image according to the target boundary frame and the category information.

Optionally, the inspecting image is subjected to feature extraction to obtain an output feature map, and the method specifically includes:

performing convolution operation on the inspection image to obtain an initial characteristic diagram;

compressing the initial characteristic diagram to obtain a compressed characteristic diagram;

performing channel-level global feature extraction on the compressed feature map, and determining a global feature weight corresponding to each channel;

and multiplying the compressed feature map and the global feature weight to obtain an output feature map.

Optionally, the location information includes: a center point coordinate, a center point offset and a target size;

the target detection of the output characteristic diagram to obtain a target detection result specifically includes:

acquiring a thermodynamic diagram of the output characteristic diagram;

acquiring the category information and the center point coordinates of the target in the inspection image based on the thermodynamic diagram;

acquiring the central point offset of a target in the inspection image according to the corresponding relationship among the category information, the category information and the central point offset;

and acquiring the target size of the target in the inspection image according to the corresponding relation among the category information, the category information and the target size.

Optionally, the decoding the position information to obtain the target bounding box specifically includes:

summing the center point coordinate and the center point offset to obtain a corrected center point coordinate;

and generating the target boundary frame according to the corrected central point coordinate and the target size.

The application also provides a detection device for the power transmission line inspection fault, which comprises:

the image acquisition unit is used for acquiring an inspection image shot when the power transmission line is inspected;

the characteristic extraction unit is used for extracting the characteristics of the patrol inspection image to obtain an output characteristic diagram;

a target detection unit, configured to perform target detection on the output feature map to obtain a target detection result, where the target detection result includes: category information and location information;

the decoding unit is used for decoding the position information to obtain a target boundary frame;

and the output unit is used for outputting the fault detection result corresponding to the inspection image according to the target boundary box and the category information.

Optionally, the feature extraction unit specifically includes:

the pre-extraction subunit is used for performing convolution operation on the inspection image to obtain an initial characteristic diagram;

the compressing subunit is used for compressing the initial characteristic diagram to obtain a compressed characteristic diagram;

the determining subunit is used for performing channel-level global feature extraction on the compressed feature map and determining a global feature weight corresponding to each channel;

and the output subunit is used for multiplying the compressed feature map and the global feature weight to obtain an output feature map.

the target detection unit specifically includes:

the first acquisition subunit is used for acquiring the thermodynamic diagram of the output characteristic diagram;

the second acquiring subunit is used for acquiring the category information and the center point coordinates of the target in the inspection image based on the thermodynamic diagram;

a third obtaining subunit, configured to obtain, according to the correspondence between the category information and the center point offset, a center point offset of the target in the inspection image;

and the fourth obtaining subunit is configured to obtain the target size of the target in the inspection image according to the corresponding relationship between the category information, and the target size.

Optionally, the decoding unit specifically includes:

the summation subunit is used for summing the center point coordinate and the center point offset to obtain a corrected center point coordinate;

and the generating subunit is used for generating the target boundary frame according to the corrected central point coordinate and the target size.

The application also provides a transmission line inspection fault detection device, the device comprises a processor and a memory:

the memory is used for storing program codes and transmitting the program codes to the processor;

the processor is used for executing the detection method for the power transmission line inspection fault according to the instructions in the program codes.

The application also provides a storage medium, which is used for storing program codes, and the program codes are used for executing the detection method for the power transmission line inspection fault.

According to the technical scheme, the method has the following advantages:

according to the detection method for the power transmission line inspection fault, the inspection image shot when the power transmission line is inspected is firstly obtained, then the inspection image is subjected to feature extraction to obtain the output feature diagram, then the output feature diagram is subjected to target detection to obtain a target detection result, and the target detection result comprises: the category information and the position information are decoded to obtain a target boundary box, and finally, a fault detection result corresponding to the inspection image is output according to the target boundary box and the category information, so that the technical problem that the real-time fault detection is difficult to realize by the conventional fault detection method is solved.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the description below are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained according to the drawings without inventive labor.

Fig. 1 is a schematic flowchart of a first embodiment of a method for detecting a power transmission line inspection fault in the embodiment of the present application;

fig. 2 is a flowchart illustrating a second embodiment of a method for detecting a power transmission line inspection fault according to the embodiment of the present application;

fig. 3 is a detection framework of a detection model of a detection method for power transmission line inspection faults in the embodiment of the present application;

fig. 4 is a schematic structural diagram of a detection apparatus for power transmission line inspection faults in the embodiment of the present application.

Detailed Description

The embodiment of the application provides a method, a device, equipment and a storage medium for detecting a power transmission line inspection fault, and solves the technical problem that the existing fault detection method is difficult to realize real-time fault detection.

In order to make the objects, features and advantages of the present invention more apparent and understandable, the following embodiments of the present invention are clearly and completely described with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the embodiments described below are only a part of the embodiments of the present invention, and not all of the embodiments of the present invention. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present application without making any creative effort belong to the protection scope of the present application.

Referring to fig. 1, fig. 1 is a schematic flowchart of a first embodiment of a method for detecting a power transmission line inspection fault according to an embodiment of the present application.

The application provides a detection method for power transmission line inspection faults, which specifically comprises the following steps:

step 101, acquiring an inspection image shot when the power transmission line is inspected.

The method can be implemented in a detection device for the power transmission line inspection fault, specifically, the device is carried on an unmanned aerial vehicle, inspection of the power transmission line is achieved through flight of the unmanned aerial vehicle, and inspection images are shot through a CCD in the flight process of the unmanned aerial vehicle.

And 102, extracting the features of the inspection image to obtain an output feature map.

And after the inspection image is obtained, performing feature extraction on the inspection image to obtain an output feature map. It is understood that the tool for extracting the features of the inspection image may be a feature extraction network, which is not specifically limited in this embodiment.

Step 103, performing target detection on the output characteristic diagram to obtain a target detection result, wherein the target detection result comprises: category information and location information.

And after the output characteristic diagram of the inspection image is obtained, carrying out target detection on the output characteristic diagram to obtain a target detection result.

Specifically, in this embodiment, the category information refers to a category to which the target belongs, such as an insulator, a power transmission tower, and the like.

And 104, decoding the position information to obtain a target boundary box.

After the target detection result is determined, the target detection result comprises the position information of the target, and the target bounding box corresponding to the target in the output image can be determined by decoding the position information.

And 105, outputting a fault detection result corresponding to the inspection image according to the target boundary frame and the category information.

After the category information and the target boundary box corresponding to the target in the inspection image are obtained, the category information and the target boundary box are displayed in an output image, and the visual loading of the target (namely, the fault) detected in the output image is completed.

It should be noted that, in this embodiment, the number of the targets in the inspection image is not limited, that is, the number of the targets in the inspection image may be one or multiple, and when the number is one, the feature extraction, the target detection, the output of the target bounding box, and the visual loading of the fault are performed according to the above method; when the number of the targets is multiple, feature extraction, target detection, target boundary box output and fault visual loading are respectively carried out for each target.

It is understood that, in this embodiment, steps 101 and 105 may be implemented by a neural network structure, that is, the input of the application network is the inspection image, and the output is the inspection image added with the object class and the object bounding box, so as to achieve end-to-end detection.

In this embodiment, the image of patrolling and examining that shoots when patrolling and examining the transmission line is obtained at first, carry out feature extraction to the image of patrolling and examining next, obtain the output characteristic map, then carry out target detection to the output characteristic map, obtain the target detection result, the target detection result includes: the method comprises the steps of obtaining a target boundary box by decoding category information and position information, outputting a fault detection result corresponding to a patrol image according to the target boundary box and the category information, and solving the technical problem that the conventional fault detection method is difficult to realize real-time fault detection.

Referring to fig. 2, fig. 2 is a schematic flowchart of a second embodiment of a method for detecting a power transmission line inspection fault in the embodiment of the present application, which may specifically include the following steps:

step 201, acquiring an inspection image shot when the power transmission line is inspected.

It should be noted that step 201 is the same as the description of step 101 in the first embodiment, and reference may be specifically made to the description of step 101, which is not described herein again.

And 202, performing convolution operation on the inspection image to obtain an initial feature map.

Specifically, the convolution operation in this embodiment may be implemented by using various network structures, such as Resnet, and the like, and a person skilled in the art may select the network structure as needed, which is not specifically limited in this embodiment.

And step 203, compressing the initial characteristic diagram to obtain a compressed characteristic diagram.

Specifically, the compression operation in this embodiment is implemented by globally averaging pooled samples.

And 204, performing channel-level global feature extraction on the compressed feature map, and determining the global feature weight corresponding to each channel.

After the compressed feature map is obtained, channel-level global feature extraction is carried out on the compressed feature map, and the global feature weight corresponding to each channel is determined.

And step 205, multiplying the compressed feature map by the global feature weight to obtain an output feature map.

After determining the compressed feature map and the global feature weight, multiplying the compressed feature map and the global feature weight to obtain an output feature map. By compression of the extrusion and attention mechanism at the channel level, efficient feature extraction is achieved.

As mentioned above, the method for detecting the power transmission line inspection fault can be implemented by a neural network, and as shown in fig. 3, a detection framework for detecting the fault by the neural network is provided.

For convenience of understanding, the method for detecting the inspection fault of the power transmission line in this embodiment is described in detail with reference to a specific implementation manner.

In this embodiment, one of the neural networks may be a cenet network, and the network for extracting features is a backbone network of the cenet network. In one implementation, the backbone network is set to Resnet.

The feature extraction network realizes a channel-level attention mechanism by adding an extrusion-excitation module in a designated convolutional layer, and realizes effective feature extraction by using deformable convolution during connection and upsampling at a designated stage, thereby completing the construction of the feature extraction network.

The extrusion-excitation module belongs to a structure for automatically learning the importance degree relation of different channel characteristics, so that the network can pay more attention to the channel characteristics with the maximum information quantity and restrain unimportant channel characteristics. The extrusion module is used for carrying out global average pooling sampling compression on the input feature map; the excitation module is used for obtaining the weights of different channels through the global features of the channel level and obtaining a final feature map by taking the product of the weights and the original feature map. The extrusion-excitation module is embedded into the feature extraction network, and channel-level optimization is carried out on the features before feature fusion, so that the extraction capability of the network on effective features is improved. After a set of convolution operations consisting of a 33 convolution, a normalized network layer and an activation function, a squeeze-excitation module consisting of an adaptive average pooling layer and two sets of convolution operations was added. In the construction process of the feature extraction network, 3 × 3 deformable convolution is used for up-sampling in each sampling layer, and 3 × 3 deformable convolution is used for interpolation projection to obtain more effective feature representation.

The deformable convolution can increase the space sampling position, adapt to scaling transformation and rotation transformation, and only increase a small amount of model complexity and calculated amount, so that the identification precision can be improved. During detection, the deformable convolution receptive field can be adaptively changed along with the size of an object, so that the object can be effectively covered by the perception of a large object, and the perception of a small object can be concentrated around the object without excessive acquisition of background information.

And step 206, acquiring a thermodynamic diagram of the output characteristic diagram.

And step 207, acquiring the category information and the center point coordinates of the target in the inspection image based on the thermodynamic diagram.

And 208, acquiring the central point offset of the target in the inspection image according to the corresponding relation among the category information, the category information and the central point offset.

And 209, acquiring the target size of the target in the inspection image according to the corresponding relation among the category information, the category information and the target size.

And after the output characteristic diagram is obtained, inputting the output characteristic diagram into the central point prediction Y branch, the central point deviation O branch and the target size S branch for detection. It can be understood that, when the network in the present application is a neural network, the three branches are implemented by a three-branch network, the three-branch network is connected to the output end of the feature extraction network, and after the output feature map output by the feature extraction network is obtained, and after the centroid coordinate prediction, the centroid offset prediction and the target size prediction are performed, 84 values of the centroid position of each target are predicted, that is, 80 centroids (i.e., centroid coordinates) of the centroid prediction Y branch result, the centroid offset O branch results x and Y, and the target size branch results w and h.

Wherein, in the central point predicting Y branch, the central point heat map is generated

I.e., a gaussian distribution map, in which the peak corresponds to the center of the target, and when the value is 1, it represents the detected center point.

In training the centroid prediction network, for each class the true frame centroid p ∈ R²First, calculate oneThe equivalent value of the low resolution is expressed as:

wherein, p represents the target center point, R represents the down-sampling multiple 4,

is equivalent to a low resolution.

And then using a Gaussian kernel, wherein the expression is as follows:

in the formula, Y_xycIs Gaussian distribution, x is the abscissa of the central point,

is the expected value of the abscissa x of the center point, y is the ordinate of the center point,

as expected value of the ordinate y of the centre point,

is the center point variance.

The center point of each real box is mapped to a heat map and pixel-level logistic regression is performed using a loss function with a balancing factor.

In the training of the centroid offset O-branch, the local offset of each centroid is a training of all target centroid offset values using L1 penalties. After training is completed, the same class shares one predicted offset value to recover discretization error caused by output step size.

The width and height of the target are regressed at the center point of the target in the target size S-branch. Assume the target bounding box of target k is

Then its corresponding center point coordinate is

The method uses a center point estimator

Generating a center point and regressing each target k to a size of

To reduce the computational burden, a single target size prediction result is used for each type of target.

And step 210, summing the central point coordinate and the central point offset to obtain a corrected central point coordinate.

And step 211, generating a target boundary frame according to the corrected central point coordinates and the target size.

The process at the point-to-target bounding box is similar to the process of decoding, having

The detected n center points of category C are aggregated. And acquiring a target bounding box by extracting 100 peak points of each category on the thermodynamic diagram, and reserving if the peak points are larger than 8 adjacent points of the peak points. Center point of target boundary box predicted by model

Deviation amount

Size of

And generating at the designated position, and directly obtaining a target boundary frame through central point estimation, so that the performance of the model can be greatly improved.

Is the coordinate of the central point, and the central point,

as an offset of the center point,

is the target width and height (target size). In the generation process of the target boundary box, firstly, the central point of model prediction is obtained

Then the offset of the training

Adding the two to obtain the shifted central point position

Finally, the size of the object obtained by the center point and the training is obtained

Generating a target bounding box, wherein the expression of the target bounding box is as follows:

and 212, outputting a fault detection result corresponding to the inspection image according to the target boundary frame and the category information.

It can be understood that the data set used for training the centret network in this embodiment is 2509 standardized drone patrol fault images covering environments such as field plains, mountains, forests, towns, etc., and the resolution is 3000 × 1700, which includes 377 insulator spontaneous explosion regions, 2655 vibration damper falling regions, and 692 bird nests. The fault images comprise self-explosion images of the insulators, falling-off images of the vibration dampers of the insulators and bird nest images. Of 3724 objects in the data set, 3665 objects account for less than 5% of the image, and the proportion of small objects is 98.42%.

The method and the device focus on small part detection, the effect of detecting the small target is obvious, the precision and the reasoning speed performance are greatly improved compared with a baseline method, and the real-time detection of the routing inspection fault of various power transmission lines is realized.

In this embodiment, the image of patrolling and examining that shoots when patrolling and examining the transmission line is obtained at first, carry out feature extraction to the image of patrolling and examining next, obtain the output characteristic map, then carry out target detection to the output characteristic map, obtain the target detection result, the target detection result includes: the category information and the position information are decoded to obtain a target boundary box, and finally, a fault detection result corresponding to the inspection image is output according to the target boundary box and the category information, so that the technical problem that the real-time fault detection is difficult to realize by the conventional fault detection method is solved.

Referring to fig. 4, fig. 4 is a schematic structural diagram of a detection apparatus for power transmission line inspection fault according to an embodiment of the present disclosure.

The embodiment of the application provides a detection device for transmission line inspection faults, and the detection device specifically comprises:

the characteristic extraction unit is used for extracting the characteristics of the patrol image to obtain an output characteristic diagram;

the target detection unit is used for carrying out target detection on the output characteristic diagram to obtain a target detection result, and the target detection result comprises: category information and location information;

and the output unit is used for outputting the fault detection result corresponding to the inspection image according to the target boundary frame and the category information.

In a specific implementation manner, the feature extraction unit specifically includes:

the determining subunit is used for performing channel-level global feature extraction on the compressed feature map and determining the global feature weight corresponding to each channel;

and the output subunit is used for multiplying the compressed feature map by the global feature weight to obtain an output feature map.

In a specific implementation, the location information includes: a center point coordinate, a center point offset and a target size;

the target detection unit specifically includes:

the third acquiring subunit is used for acquiring the central point offset of the target in the inspection image according to the corresponding relationship among the category information, the category information and the central point offset;

and the fourth acquiring subunit is used for acquiring the target size of the target in the inspection image according to the corresponding relation among the category information, the category information and the target size.

In a specific implementation manner, the decoding unit specifically includes:

and the generating subunit is used for generating a target boundary frame according to the corrected central point coordinate and the target size.

The embodiment of the application further provides a detection device for the inspection fault of the power transmission line, and the device comprises a processor and a memory: the memory is used for storing the program codes and transmitting the program codes to the processor; the processor is used for executing the detection method for the power transmission line inspection fault according to the instructions in the program codes.

The embodiment of the application also provides a storage medium, wherein the storage medium is used for storing program codes, and the program codes are used for executing the detection method for the power transmission line inspection fault in the embodiment of the application.

It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.

The embodiments in the present specification are all described in a progressive manner, and each embodiment focuses on differences from other embodiments, and portions that are the same and similar between the embodiments may be referred to each other.

As will be appreciated by one of skill in the art, embodiments of the present application may be provided as a method, apparatus, or computer program product. Accordingly, embodiments of the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, embodiments of the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and so forth) having computer-usable program code embodied therein.

Embodiments of the present application are described with reference to flowchart illustrations and/or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing terminal to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing terminal to cause a series of operational steps to be performed on the computer or other programmable terminal to produce a computer implemented process such that the instructions which execute on the computer or other programmable terminal provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

While preferred embodiments of the present application have been described, additional variations and modifications of these embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including the preferred embodiment and all changes and modifications that fall within the true scope of the embodiments of the present application.

Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or terminal. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in the process, method, article, or terminal equipment comprising the element.

The above embodiments are only used to illustrate the technical solutions of the present application, and not to limit the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions in the embodiments of the present application.

Claims

1. A detection method for power transmission line inspection faults is characterized by comprising the following steps:

acquiring a patrol image shot when the power transmission line is patrolled;

decoding the position information to obtain a target boundary box;

2. The method for detecting the power transmission line inspection fault according to claim 1, wherein the step of performing feature extraction on the inspection image to obtain an output feature map specifically comprises the following steps:

3. The method for detecting the power transmission line inspection fault according to claim 1, wherein the position information comprises: a center point coordinate, a center point offset and a target size;

the target detection is performed on the output characteristic diagram to obtain a target detection result, and the method specifically includes:

acquiring a thermodynamic diagram of the output characteristic diagram;

acquiring the central point offset of the target in the inspection image according to the corresponding relation among the category information, the category information and the central point offset;

4. The method for detecting the inspection fault of the power transmission line according to claim 3, wherein the decoding the position information to obtain the target bounding box specifically comprises:

5. The utility model provides a transmission line patrols and examines detection device of trouble which characterized in that includes:

the characteristic extraction unit is used for extracting the characteristics of the inspection image to obtain an output characteristic diagram;

6. The device for detecting the power transmission line inspection fault according to claim 5, wherein the feature extraction unit specifically comprises:

the determining subunit is configured to perform channel-level global feature extraction on the compressed feature map, and determine a global feature weight corresponding to each channel;

7. The apparatus of claim 5, wherein the location information includes: a center point coordinate, a center point offset and a target size;

the target detection unit specifically includes:

the third acquiring subunit is configured to acquire a center point offset of the target in the inspection image according to the corresponding relationship between the category information and the category information;

8. The device for detecting the power transmission line inspection fault according to claim 7, wherein the decoding unit specifically comprises:

9. The utility model provides a transmission line patrols and examines check out detection equipment of trouble which characterized in that, equipment includes treater and memory:

the processor is used for executing the detection method for the power transmission line inspection fault according to the instructions in the program codes, wherein the method is as defined in any one of claims 1 to 4.

10. A storage medium for storing program code for performing the method of transmission line inspection fault detection according to any one of claims 1 to 4.