CN115937492A - Transformer equipment infrared image identification method based on feature identification - Google Patents
Transformer equipment infrared image identification method based on feature identification Download PDFInfo
- Publication number
- CN115937492A CN115937492A CN202211590327.XA CN202211590327A CN115937492A CN 115937492 A CN115937492 A CN 115937492A CN 202211590327 A CN202211590327 A CN 202211590327A CN 115937492 A CN115937492 A CN 115937492A
- Authority
- CN
- China
- Prior art keywords
- feature
- power transformation
- infrared image
- transformation equipment
- equipment
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Abstract
The invention discloses a transformer equipment infrared image recognition method based on feature recognition, which aims to solve the problem that the transformer equipment target is difficult to detect under the condition that equipment and a background in an infrared image are difficult to distinguish. The method utilizes the priori knowledge to position and classify various power transformation equipment, solves the problem that the target of the power transformation equipment is difficult to detect under the condition that the equipment and the background in the infrared image are difficult to distinguish, has more accurate detection and stronger universality, and does not need manual participation in the detection process.
Description
Technical Field
The invention relates to the field of visual image recognition, in particular to a transformer equipment infrared image recognition method based on feature recognition.
Background
Temperature monitoring and thermal fault diagnosis of the power transformation equipment are important for safe operation of the power transformation equipment. In recent years, operation and maintenance personnel use thermal infrared imagers mounted on monitoring platforms such as inspection robots and unmanned aerial vehicles to carry out intelligent inspection on substation equipment, so that the workload of manually acquiring infrared images of the substation equipment is reduced, but the identification and detection of mass infrared images of the substation equipment also depend on a manual method which is long in time consumption, complex in process and low in efficiency. Therefore, it is very necessary to provide an efficient method for automatically detecting infrared images of power transformation equipment.
In recent years, the development of artificial intelligence enables more and more target detection methods based on deep learning to be applied to infrared image detection of electrical equipment. The current mainstream target detection deep learning algorithm can be divided into two major categories, namely a two-stage detection algorithm and a single-stage detection algorithm, the first category of method is carried out in two steps, firstly, a candidate region is generated by using structures such as a region suggestion network (RPN) and the like, and then, target classification and position adjustment are carried out on the candidate region. A representative algorithm is Faster R-CNN. The second method does not need to generate a candidate region independently, and classification and frame regression parameters of the target can be obtained only through a full-convolution structure. Typical methods are YOLO, SSD, etc. Although the detection speed of the two-stage algorithm is lower than that of the single-stage algorithm, the accuracy is obviously improved, and the method is widely applied to the field of electric power.
The complex background of the transformer substation is an important factor influencing the automatic detection precision of the infrared image of the transformer equipment. Due to the fact that temperature difference between the power transformation equipment and the background is small, and due to interference of factors such as clutter and noise caused by extreme weather, the target detection algorithm is difficult to distinguish and detect the power transformation equipment and the background environment. Meanwhile, the conventional detection method generally aims at the implementation mode of a target detection algorithm on a public data set, lacks of pertinence analysis on infrared images of the power transformation equipment, and is difficult to detect a plurality of power transformation equipment with high similarity at the same time, so that the detection accuracy of the conventional method is low.
Disclosure of Invention
The invention aims to provide a transformer equipment infrared image identification method based on feature identification, which is used for positioning and classifying various transformer equipment by using interest features in an infrared image containing various transformer equipment, so that the problem that the transformer equipment target is difficult to detect under the condition that the equipment and the background in the infrared image are difficult to distinguish is solved.
The invention is realized by the following technical scheme:
a transformer equipment infrared image identification method based on feature identification comprises the following steps:
1. a transformer equipment infrared image identification method based on feature identification is characterized by comprising the following steps:
s1, acquiring infrared image data of various power transformation equipment, and performing primary type division on the power transformation equipment in the infrared image data to obtain a type set;
s2, detecting interest characteristics of different power transformation equipment, generating a characteristic group, and extracting characteristics corresponding to the power transformation equipment which belongs to the classification under the type set in the training set; respectively carrying out primary labeling on different power transformation equipment, and dividing a training set and a verification set according to the result of the primary labeling;
s3, after the extracted features are subjected to local processing, inputting the local image information of the power transformation equipment containing the extracted features in the training set into a learning network, starting recognition training on the training set by using the extracted features as training parameters of the learning network, and outputting a type recognition model containing a weight file;
s4: and identifying the power transformation equipment based on the type identification model.
As an optional mode of the present invention, the interest feature is a visually significant feature of the power transformation device, and a feature group consisting of a plurality of different kinds of visually significant features is generated according to a difference between visually significant features of different power transformation devices; and respectively extracting the visually significant features of the power transformation equipment according to different types contained in the feature groups.
As an optional aspect of the present invention, in step S2, after performing feature extraction on the power transformation device, a first feature set and a second feature set are generated; the first characteristic set is composed of visual salient features corresponding to the power transformation equipment, and the second characteristic set is composed of image information of the power transformation equipment after the features are extracted.
As an optional mode of the present invention, in step S2, labelMe is adopted to perform preliminary labeling on the infrared image data, where the preliminary labeling includes an overall labeling and a feature labeling, the feature labeling is to label visual features of the power transformation equipment according to a first feature set to generate a first label set, and the overall labeling is to label the infrared image data of the power transformation equipment according to a second feature set to generate a second label set; and determining the capacity of the type set according to the different times and types of the primary annotation, and integrating the first annotation set and the second annotation set to manufacture a data set.
As an optional mode of the present invention, before the step S3, the method further includes randomly ordering the training set, and sending the randomly ordered pictures in the training set to the feature extraction backbone network in batches in order of the first labeled set for feature extraction, so as to generate a multi-level feature map. As an optional mode of the invention, the feature map is input into the regional suggestion network and then is subjected to convolution processing, and concentrated feature information is output; generating at least one reference frame on the feature map via the full convolution network; and selecting a target candidate region of each real frame for different pictures according to the intersection ratio of the reference frames by taking any reference frame as a starting point, and calculating the regression loss and the classification loss of the region suggestion network according to the candidate regions corresponding to the target candidate frame and the reference frame.
As an optional mode of the invention, the feature map is sent to a pooling layer of the target candidate region, and feature maps with consistent sizes are output; carrying out classification calculation on the feature graph through full connection and a classification layer, outputting probability vectors of the feature graph and the type set, then carrying out multiple times of training on the training set according to a preset training step length, and outputting and storing a weight file of the type recognition model after the training is finished; and inputting the test picture into a model with a weight file for detection, and outputting a detection result.
As an alternative of the present invention, the determination method of the visually significant features is:
after the interest features of the power transformation equipment are extracted, identifying the local images of the power transformation equipment after the interest features are extracted, and if the type set to which the local images of the power transformation equipment belong cannot be identified, considering the interest features as the visual salient features of the power transformation equipment.
In addition, in order to achieve the above object, the present embodiment further provides an electronic device, which includes a memory and a processor, where the memory stores a computer program that is executable on the processor, and the processor implements the steps of the method for identifying an infrared image of a power transformation device based on feature identification when executing the computer program.
In addition, to achieve the above object, the present embodiment further provides a storage medium having a computer program stored thereon, where the computer program is executed by a processor to implement the steps in the above-mentioned transformer equipment infrared image recognition method based on feature recognition.
Compared with the prior art, the invention has the following advantages and beneficial effects:
according to the target detection method based on deep learning, the problem that the power transformation equipment is difficult to position and classify in the complex infrared image is solved, the visual detection precision of different power transformation equipment is improved, the method universality is strong, and the method design has inspiration on related problems.
Drawings
The accompanying drawings, which are included to provide a further understanding of the embodiments of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the principles of the invention. In the drawings:
fig. 1 is a schematic flow chart illustrating steps of a method for identifying infrared image types of power transformation equipment according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of a type recognition model according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to examples and accompanying drawings, and the exemplary embodiments and descriptions thereof are only used for explaining the present invention and are not meant to limit the present invention.
Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, 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, 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 the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (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 apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, 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 apparatus 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.
Examples
Referring to fig. 1-2, the present embodiment provides a transformer equipment infrared image recognition method based on feature recognition, and aims to solve the problem that it is difficult to detect a transformer equipment target under the condition that equipment is difficult to distinguish from a background in an infrared image. Which comprises the following steps:
s1: acquiring infrared image data of various power transformation equipment, and performing primary type division on the power transformation equipment in the infrared image data to obtain a type set; the type set can include voltage transformer, current transformer, lightning arrester and other common substation equipment according to the difference of actual application environment.
S2: and detecting interest characteristics of different power transformation equipment, generating a characteristic group, and extracting characteristics corresponding to the power transformation equipment of which the training set belongs to the classification under the type set. In this embodiment, the interesting features are visually significant features of the power transformation device, such as: the grading ring, the expansion chamber shell and the top oil tank shell on the top of the equipment can assist in confirming a 220kV zinc oxide lightning arrester, a 220kV current transformer and a 220kV voltage transformer respectively. Generating a feature group consisting of a plurality of different types of visual salient features according to different visual salient features of different power transformation equipment; and respectively extracting visual salient features of the power transformation equipment according to different types contained in the feature groups, then respectively carrying out primary labeling on different power transformation equipment, and dividing a training set and a verification set according to the result of the primary labeling.
In step S2, the manner of determining the visually significant features is: after the interest features of the power transformation equipment are extracted, identifying the local images of the power transformation equipment with the extracted interest features, and if the type set to which the local images belong cannot be identified, considering the interest features as the visual salient features of the power transformation equipment. For example, after an infrared image of the device is partially shielded, a person or another existing visual recognition device/system tries to distinguish the device, after a certain part of the device is shielded, the situation that the kind of the person or another existing visual recognition device/system is difficult to distinguish occurs, and when only the shielded part of the image is displayed, the person or another existing visual recognition device/system can accurately distinguish the kind of the device, which indicates that the shielded part plays an important role in recognition and classification, and the visual phenomenon is combined with domain knowledge of the power transformation device, and the visual phenomenon is used as a significant visual feature of the power transformation device, and the feature is fused into the training work of a subsequent deep learning model.
And after the extracted features are subjected to local processing, merging and storing the processed features. And inputting the local image information of the power transformation equipment containing the extracted features in the training set into the type recognition model. In this embodiment, labelMe software is used for performing preliminary labeling on infrared image data, the preliminary labeling includes whole labeling and feature labeling, and the feature labeling is to label visual features of the power transformation equipment according to a first feature set to generate a first label set, that is, a significant visual feature set of different power transformation equipment. And the integral labeling is to label the infrared image data of the power transformation equipment according to a second characteristic set to generate a second label set, namely the local infrared image of the power transformation equipment which cannot be directly identified after the visual salient features are shielded. And determining the capacity of the type set according to the different times and types of the primary annotation, and integrating the first annotation set and the second annotation set to manufacture a data set.
S3: and after the extracted features are subjected to local processing, inputting the local image information of the power transformation equipment containing the extracted features in the training set into a learning network, starting recognition training on the training set by using the extracted features as training parameters of the learning network, and outputting a type recognition model containing a weight file.
Before the step S3, randomly ordering the training set, and sending the randomly ordered pictures in the training set to the feature extraction backbone network in batches in order of the first labeled set to perform feature extraction, so as to generate a multi-level feature map. Inputting the feature map into the regional suggestion network, performing convolution processing, and outputting concentrated feature information; at least one reference box is generated on the feature map over the full convolution network. And selecting a target candidate region of each real frame for different pictures according to the intersection ratio of the reference frames by taking any reference frame as a starting point, and calculating the regression loss and the classification loss of the region suggestion network according to the candidate regions corresponding to the target candidate frame and the reference frame.
Then, the feature maps are sent to a pooling layer of the target candidate area, and feature maps with consistent sizes are output; and carrying out classification calculation on the feature map through full connection and a classification layer, outputting the probability vectors of the feature map and the type set to which the feature map belongs, and further optimizing by utilizing position regression of the regional suggestion network. And repeating the training steps, training the training set for multiple times according to the preset training step length to obtain an accurate target frame, and accurately identifying the power transformation equipment. Outputting and storing the weight file of the type recognition model after the training is finished; and inputting the test picture into a model with a weight file for detection, and outputting a detection result.
Referring to fig. 2 again, fig. 2 is a structural diagram of the type recognition model in the present embodiment, and by integrating the above methods and steps, for any image with a size of P × Q, the image is first scaled to a fixed size of M × N, and then the M × N image is sent to the network; firstly, feature extraction is performed through a feature extraction network. And then inputting a region suggestion network into the feature map, and outputting a series of candidate regions and classification probabilities thereof. And finally, the feature map enters a detection network, the detection network comprises a candidate region pooling layer and a classification layer, the detection network calculates the target confidence of 3 kinds of power transformation equipment in each region through a full connection layer and the classification layer by using the obtained region suggestions, and the label with the highest confidence is the classification result.
The regional suggestion network has two branches, the first branch is responsible for pixel classification so as to obtain the foreground and the background; the second branch duplicates the edge regression, calculates the bounding box offset to determine the position of the candidate box, and the specific implementation is as follows: the bounding box to be determined is an [ x, y, w, h ]]Coordinate values of the form. The real selection area is G, the original selection area is A, the regression window is G ', G' is a regression window very close to the real window G, so a transformation F is sought, and A [ x, y, w, h ]]Conversion to G' [ G ] x ,G y ,G w ,G h ]The idea is to perform a translation transformation to approximate the center point, and then perform a scaling to approximate the size, thereby obtaining the result
G′ x ＝A w ·d x (A)+A x
G′ y ＝A h ·d y (A)+A y
Through the four transformations, the G' coordinate can be obtained, and the network needs to learn d in this embodiment x (A)d y (A)d w (A)d h (A) These four transformations.
In addition, in order to achieve the above object, the present embodiment further provides an electronic device, which includes a memory and a processor, where the memory stores a computer program that is executable on the processor, and the processor implements the steps of the method for identifying an infrared image of a power transformation device based on feature identification when executing the computer program.
In addition, to achieve the above object, the present embodiment further provides a storage medium having a computer program stored thereon, where the computer program is executed by a processor to implement the steps in the above-mentioned transformer equipment infrared image recognition method based on feature recognition.
Therefore, in the infrared image containing various power transformation devices, the prior knowledge is utilized to position and classify the various power transformation devices, and the problem that the target of the power transformation device is difficult to detect under the condition that the device and the background in the infrared image are difficult to distinguish is solved.
The above-mentioned embodiments are intended to illustrate the objects, technical solutions and advantages of the present invention in further detail, and it should be understood that the above-mentioned embodiments are merely exemplary embodiments of the present invention, and are not intended to limit the scope of the present invention, and any modifications, equivalent substitutions, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (10)
1. A transformer equipment infrared image identification method based on feature identification is characterized by comprising the following steps:
s1, acquiring infrared image data of various power transformation equipment, and performing primary type division on the power transformation equipment in the infrared image data to obtain a type set;
s2, detecting interest characteristics of different power transformation equipment, generating a characteristic group, and extracting characteristics corresponding to the power transformation equipment which belongs to the classification under the type set in the training set; respectively carrying out primary labeling on different power transformation equipment, and dividing a training set and a verification set according to the result of the primary labeling;
s3, after the extracted features are subjected to local processing, inputting the local image information of the power transformation equipment containing the extracted features in the training set into a learning network, starting recognition training on the training set by using the extracted features as training parameters of the learning network, and outputting a type recognition model containing a weight file;
s4: and identifying the power transformation equipment based on the type identification model.
2. The infrared image recognition method for the power transformation equipment based on the feature recognition is characterized in that the interest features are visual salient features of the power transformation equipment, and a feature group consisting of a plurality of different types of visual salient features is generated according to different visual salient features of different power transformation equipment; and respectively extracting the visually significant features of the power transformation equipment according to different types contained in the feature groups.
3. The transformer equipment infrared image recognition method based on the feature recognition is characterized in that in the step S2, after the feature extraction is carried out on the transformer equipment, a first feature set and a second feature set are generated; the first feature set is composed of visual salient features corresponding to the power transformation equipment, and the second feature set is composed of image information of the power transformation equipment after the features are extracted.
4. A power transformation equipment infrared image identification method based on feature identification according to claim 3, characterized in that in step S2, labelMe is adopted to perform preliminary labeling on infrared image data, the preliminary labeling includes an overall labeling and a feature labeling, the feature labeling is to label visual features of power transformation equipment according to the first feature set to generate a first label set, and the overall labeling is to label infrared image data of power transformation equipment according to the second feature set to generate a second label set; and determining the capacity of the type set according to different times and types of the preliminary annotation, and integrating the first annotation set and the second annotation set to manufacture a data set.
5. A transformer equipment infrared image recognition method based on feature recognition as claimed in claim 4, characterized by further comprising the sub-steps, before the step S3, of: and randomly ordering the training set, and sending the pictures in the randomly ordered training set into a feature extraction backbone network in batches by taking the first labeled set as a sequence to perform feature extraction so as to generate a multi-stage feature map.
6. The transformer equipment infrared image recognition method based on feature recognition is characterized in that after the feature map is input into the area suggestion network, convolution processing is carried out, and concentrated feature information is output; generating at least one reference frame on the feature map through the full convolution network; and selecting a target candidate region of each real frame for different pictures according to the intersection ratio of the reference frames by taking any reference frame as a starting point, and calculating the regression loss and the classification loss of the region suggestion network according to the candidate regions corresponding to the target candidate frame and the reference frame.
7. The infrared image recognition method for the power transformation equipment based on the feature recognition is characterized in that the feature map is sent into a pooling layer of a target candidate area, and feature maps with the same size are output; carrying out classification calculation on the feature graph through full connection and a classification layer, outputting probability vectors of the feature graph and the type set, then carrying out multiple times of training on the training set according to a preset training step length, and outputting and storing a weight file of the type recognition model after the training is finished; and inputting the test picture into a model with a weight file for detection, and outputting a detection result.
8. The transformer equipment infrared image recognition method based on feature recognition is characterized in that the visually significant features are determined in the following manner:
after the interest features of the power transformation equipment are extracted, identifying the local images of the power transformation equipment with the extracted interest features, and if the type set to which the local images belong cannot be identified, considering the interest features as the visual salient features of the power transformation equipment.
9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program operable on the processor, and wherein the processor implements the steps of the method for transformer equipment infrared image recognition based on feature recognition according to any one of claims 1 to 8 when executing the computer program.
10. A storage medium having a computer program stored thereon, wherein the computer program is configured to implement the steps of the method for identifying an infrared image of a power transformation device based on feature identification according to any one of claims 1 to 8 when executed by a processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211590327.XA CN115937492A (en) | 2022-12-12 | 2022-12-12 | Transformer equipment infrared image identification method based on feature identification |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211590327.XA CN115937492A (en) | 2022-12-12 | 2022-12-12 | Transformer equipment infrared image identification method based on feature identification |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115937492A true CN115937492A (en) | 2023-04-07 |
Family
ID=86653697
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211590327.XA Pending CN115937492A (en) | 2022-12-12 | 2022-12-12 | Transformer equipment infrared image identification method based on feature identification |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115937492A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117705816A (en) * | 2024-02-06 | 2024-03-15 | 西南交通大学 | Unmanned aerial vehicle-based steel rail surface defect detection method, system, equipment and medium |
| CN117705816B (en) * | 2024-02-06 | 2024-04-26 | 西南交通大学 | Unmanned aerial vehicle-based steel rail surface defect detection method, system, equipment and medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200184252A1 (en) * | 2018-12-10 | 2020-06-11 | International Business Machines Corporation | Deep Learning Network for Salient Region Identification in Images |
| US20210326639A1 (en) * | 2020-10-23 | 2021-10-21 | Beijing Baidu Netcom Science and Technology Co., Ltd | Image recognition method, electronic device and storage medium |
| CN114596244A (en) * | 2020-12-04 | 2022-06-07 | 国网山东省电力公司济南供电公司 | Infrared image identification method and system based on visual processing and multi-feature fusion |
-
2022
- 2022-12-12 CN CN202211590327.XA patent/CN115937492A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200184252A1 (en) * | 2018-12-10 | 2020-06-11 | International Business Machines Corporation | Deep Learning Network for Salient Region Identification in Images |
| US20210326639A1 (en) * | 2020-10-23 | 2021-10-21 | Beijing Baidu Netcom Science and Technology Co., Ltd | Image recognition method, electronic device and storage medium |
| CN114596244A (en) * | 2020-12-04 | 2022-06-07 | 国网山东省电力公司济南供电公司 | Infrared image identification method and system based on visual processing and multi-feature fusion |
Non-Patent Citations (2)
| Title |
|---|
| 1: "1", 1, vol. 1, no. 1, pages 1 * |
| 李文璞;谢可;廖逍;李小宁;王皓;: "基于Faster RCNN变电设备红外图像缺陷识别方法", 南方电网技术, no. 12, pages 79 - 83 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117705816A (en) * | 2024-02-06 | 2024-03-15 | 西南交通大学 | Unmanned aerial vehicle-based steel rail surface defect detection method, system, equipment and medium |
| CN117705816B (en) * | 2024-02-06 | 2024-04-26 | 西南交通大学 | Unmanned aerial vehicle-based steel rail surface defect detection method, system, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109543606B (en) | Human face recognition method with attention mechanism | |
| CN111784685B (en) | Power transmission line defect image identification method based on cloud edge cooperative detection | |
| CN107742093B (en) | Real-time detection method, server and system for infrared image power equipment components | |
| CN111047554B (en) | Composite insulator overheating defect detection method based on instance segmentation | |
| CN111798456A (en) | Instance segmentation model training method and device and instance segmentation method | |
| CN112837315B (en) | Deep learning-based transmission line insulator defect detection method | |
| CN110133443B (en) | Power transmission line component detection method, system and device based on parallel vision | |
| CN110852327A (en) | Image processing method, image processing device, electronic equipment and storage medium | |
| CN115424017B (en) | Building inner and outer contour segmentation method, device and storage medium | |
| CN114140665A (en) | Dense small target detection method based on improved YOLOv5 | |
| CN111027538A (en) | Container detection method based on instance segmentation model | |
| CN113850136A (en) | Yolov5 and BCNN-based vehicle orientation identification method and system | |
| CN116579992A (en) | Small target bolt defect detection method for unmanned aerial vehicle inspection | |
| CN111539456A (en) | Target identification method and device | |
| CN111241905A (en) | Power transmission line nest detection method based on improved SSD algorithm | |
| CN116452604A (en) | Complex substation scene segmentation method, device and storage medium | |
| CN116452850A (en) | Road ponding area identification method based on data mining and deep learning | |
| CN115937492A (en) | Transformer equipment infrared image identification method based on feature identification | |
| CN114332814A (en) | Parking frame identification method and device, electronic equipment and storage medium | |
| CN115100546A (en) | Mobile-based small target defect identification method and system for power equipment | |
| CN114639013A (en) | Remote sensing image airplane target detection and identification method based on improved Orient RCNN model | |
| CN114445691A (en) | Model training method and device, electronic equipment and storage medium | |
| CN113239931A (en) | Logistics station license plate recognition method | |
| CN112614094A (en) | Insulator string abnormity positioning and identifying method based on sequence state coding | |
| CN112329863B (en) | Method for identifying state of isolation switch in traction substation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |