CN113496210A - Attention mechanism-based photovoltaic string tracking and fault tracking method - Google Patents
Attention mechanism-based photovoltaic string tracking and fault tracking method Download PDFInfo
- Publication number
- CN113496210A CN113496210A CN202110687502.6A CN202110687502A CN113496210A CN 113496210 A CN113496210 A CN 113496210A CN 202110687502 A CN202110687502 A CN 202110687502A CN 113496210 A CN113496210 A CN 113496210A
- Authority
- CN
- China
- Prior art keywords
- fault
- string
- tracking
- feature map
- feature
- 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.)
- Granted
Links
- 230000007246 mechanism Effects 0.000 title claims abstract description 48
- 238000000034 method Methods 0.000 title claims abstract description 46
- 230000004927 fusion Effects 0.000 claims abstract description 33
- 230000008569 process Effects 0.000 claims abstract description 17
- 238000000605 extraction Methods 0.000 claims abstract description 10
- 238000010586 diagram Methods 0.000 claims description 54
- 239000013598 vector Substances 0.000 claims description 11
- 238000012549 training Methods 0.000 claims description 6
- 230000009471 action Effects 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 230000010534 mechanism of action Effects 0.000 claims description 3
- 238000001514 detection method Methods 0.000 abstract description 8
- 238000013527 convolutional neural network Methods 0.000 description 15
- 238000012423 maintenance Methods 0.000 description 6
- 230000009466 transformation Effects 0.000 description 4
- 238000000844 transformation Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000010248 power generation Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 208000033793 Neuroendocrine tumor of stomach Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
- 208000036506 well differentiated low or intermediate grade gastric neuroendocrine tumor Diseases 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a photovoltaic string tracking and fault tracking method based on an attention mechanism, which comprises the steps of inputting a target frame into a pre-trained CNN network tracking model, outputting target tracking information of a photovoltaic string and numbering the target tracking information, and outputting fault tracking information in the string, wherein the CNN network tracking model comprises three modules of feature extraction, feature fusion and prediction, and a feature extraction part utilizes CNN high-dimensional and low-dimensional information to assist in feature extraction; the feature fusion component adds an attention channel and a generalized general GNet channel. Because the high-dimensional and low-dimensional features are added into the feature extraction channel, the robustness in the motion process is improved; meanwhile, as the GNet channel, the context enhancement module and the cross feature enhancement module are introduced into the feature fusion part, semantic features are fully considered in the tracking process, the robustness of attributes such as motion change and the like in the tracking process is improved, and more accurate results are obtained when photovoltaic string tracking and fault detection tracking are carried out.
Description
Technical Field
The invention belongs to the technical field of computer digital image processing, and particularly relates to a photovoltaic string tracking and fault tracking method based on an attention mechanism.
Background
At present, photovoltaic power generation is a representative technology in the field of new energy, and is widely applied at home and abroad. In consideration of environmental differences of distribution of the photovoltaic installation machines in China, the photovoltaic power generation panel often causes hot spots due to external factors such as external weather and environment, and the hot spots can cause local power transmission faults, affect power transmission efficiency and even burn out solar components.
With the development of image processing technology, tracking and detection technology based on deep learning is performed better and better. Compared with the traditional method, the photovoltaic string tracking and fault tracking method based on deep learning applies target tracking to the photovoltaic field, improves photovoltaic management and fault detection efficiency, and facilitates personnel maintenance.
For existing fault detection and photovoltaic management work, infrared image fault detection or a traditional manual inspection method is mostly adopted, and high efficiency and high accuracy cannot be simultaneously met for large-scale panel management and fault inspection work, so that photovoltaic equipment cannot be timely managed, and serious energy waste and potential safety hazards are caused.
Disclosure of Invention
The invention aims to provide a photovoltaic string tracking and fault tracking method based on an attention mechanism, which can accurately describe strings and fault information in the strings and improve the detection accuracy.
The technical scheme adopted by the invention is that the attention mechanism-based photovoltaic string tracking and fault tracking method is implemented according to the following steps:
step 4, inputting the denoised string feature map and the enhanced and fused fault feature map into a cross feature enhancement module CFA, carrying out cross fusion based on attention, and outputting a fused feature map;
and 6, outputting the fault type and position of the photovoltaic string according to the thermodynamic diagram and the classification result.
The invention is also characterized in that:
the specific process of the step 2 is as follows:
the method comprises the steps that a video of a photovoltaic string is collected by an unmanned aerial vehicle to serve as a target video, the target video is input into a trained CNN network tracking model based on an attention mechanism, string features are extracted through a high-dimensional feature extraction channel con-5 in the CNN network based on the attention mechanism to obtain a string feature map, and target image features are selected through a low-dimensional channel con-4 to extract internal fault detail information of the string to obtain a fault feature map.
XEC=X+MultiHead(X+Px,X+Px,X) (1)
wherein X represents the input of ECA, pxFor spatial position coding, XECIs the output of the ECA.
In step 4, space position coding is adopted in the cross feature enhancement module CFA, and an FFN module is introduced in the cross feature enhancement module CFA, wherein the FFN module is a fully-connected feedforward network consisting of two linear transformations, and a ReLU is arranged between the two linear transformations, namely
FFN(x)=max(0,xW1+b1)W2+b2 (2)
Where the symbol w denotes the weight matrix, b denotes the basis vectors and the subscripts denote the different layers of the fusion.
The specific process of the step 4 is as follows: inputting the denoised string feature map and the enhanced and fused fault feature map into a cross feature enhancement module CFA, repeating the fusion part for N times under the action mechanism of the CFA, and outputting the fused feature map;
among them, the CFA mechanism of action is summarized as:
wherein, XqIs a vector representation form, P, of the enhanced fused fault characteristic diagramqIs corresponding to XqSpatial position coding of, XkvIs a denoised string feature map vector representation form, pkvIs XkvThe spatial encoding of the coordinates is performed,is output by fusion of multi-headed intersection features, XCFIs the output result of the CFA.
And in the step 5, the prediction head comprises a classification branch and a regression branch, and the classification branch and the regression branch respectively comprise three layers of perceptrons with hidden dimensions d and ReLU activation functions.
The specific process of the step 5 is as follows: inputting the fused feature map into a prediction head, classifying through binary cross entropy loss in a classification branch, and performing regression training in a regression branch to obtain a fault foreground/background classification result;
wherein the classification branch is defined as:
Lcls=-∑[yilog(pj)+(1-yi)log(1-pj)] (4)
yjgroup-truth tag, y, representing the jth sample j1 denotes foreground, pjRepresenting the probability of belonging to the foreground predicted by the learning model;
the regression loss in the regression branch is defined as:
The specific process of the step 6 is as follows: obtaining a photovoltaic string prediction position according to the thermodynamic diagram, and tracking the serial number of the string according to the photovoltaic string prediction position; and simultaneously, obtaining a fault foreground/background classification result according to the step 5, and outputting a classification and tracking result of the faults in the string.
The invention has the beneficial effects that:
according to the photovoltaic string tracking and fault tracking method based on the attention mechanism, the serial number tracking of the photovoltaic string is completed through the characteristics of different levels of a pre-trained CNN tracking network based on the attention mechanism in combination with the characteristics fusion of the attention mechanism, and the fault tracking output of the inside of the string is completed at the same time, so that the string and the fault information inside the string can be accurately described, the detection accuracy is improved, the operation and maintenance cost is reduced, and the maintenance of workers is facilitated.
Drawings
FIG. 1 is a schematic diagram of an overall network structure of a photovoltaic string tracking and fault tracking method based on an attention mechanism according to the present invention;
FIG. 2 is a schematic flow chart of a photovoltaic string tracking and fault tracking method based on an attention mechanism according to the present invention;
FIG. 3 is a diagram of a feature fusion section context enhancement module (ECA) provided by the present invention;
FIG. 4 is a schematic diagram of a feature fusion part cross feature enhancement module (CFA) provided by the present invention;
FIG. 5 is an original image of a photovoltaic string collected in an embodiment of the present invention;
FIG. 6 is a diagram illustrating the effect of tracking group string numbers according to an embodiment of the present invention;
fig. 7 is a string numbering and fault location diagram of a photovoltaic string tracking and fault tracking method based on an attention mechanism according to an embodiment of the present invention.
Detailed Description
The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.
The invention provides a photovoltaic string tracking and fault tracking method based on an attention mechanism, wherein a network of the method mainly comprises three parts of feature extraction, feature fusion and target prediction, can accurately describe a string and fault information in the string, and realizes serial number tracking of the photovoltaic string and accurate output of the fault information, as shown in figure 1.
The characteristic extraction part is combined with the pre-trained attention-based CNN characteristics of different levels, wherein the high dimension contains more semantic information and can be used for extracting a string characteristic diagram; the lower dimension contains more local features that will help to extract the fault from the string.
And a feature fusion part combines an attention mechanism, and firstly utilizes two context enhancement modules (ECA) to respectively perform feature enhancement on the string and the fault feature map so as to respectively enhance the string and the fault feature. And then, carrying out ECA enhancement by using a cross feature enhancement (CFA) module, and carrying out weighted fusion on the group string feature diagram and the fault feature diagram in multiple scales to enhance the representation capability of the feature diagram. Sending the group string characteristics into a GNet network for generating a group string thermodynamic diagram; for the fault characteristics, a cross characteristic enhancement (CFA) module is utilized to perform weighted fusion on the string characteristic diagram and the fault characteristic diagram in multiple scales so as to enhance the representation capability of the characteristic diagram.
The prediction unit obtains a predicted string position based on the string thermodynamic diagram, and performs a tracking number of the string. And for the fault fusion characteristic diagram, a prediction head comprising a classification branch and a prediction branch is adopted to classify and regress the characteristic diagram, so that a foreground/background classification result and normalized coordinates of a fault prediction area are obtained, and the fault positioning output work is completed.
The invention relates to a photovoltaic string tracking and fault tracking method based on an attention mechanism, which is implemented according to the following steps as shown in figure 2:
collecting a video of a photovoltaic string as a target video by an unmanned aerial vehicle, inputting the target video into a trained CNN network tracking model based on an attention mechanism, extracting string characteristics by a high-dimensional characteristic extraction channel con-5(CNN 15 th layer) in the CNN network based on the attention mechanism, and obtaining a string characteristic diagram fzTarget image feature selection is carried out through a low-dimensional channel con-4(CNN 10 th layer) to extract fault detail information in a group string, faults with similar appearances can be better distinguished, and a fault feature map f is obtainedx。
step 4, inputting the denoised string feature map and the enhanced and fused fault feature map into a cross feature enhancement module CFA, carrying out cross fusion based on attention, and outputting a fused feature map;
XEC=X+MultiHead(X+Px,X+Px,X) (1)
wherein X represents the input of ECA, pxFor spatial position coding, XECIs the output of the ECA.
In step 4, space position coding is adopted in the cross feature enhancement module CFA, and an FFN module is introduced in the cross feature enhancement module CFA, wherein the FFN module is a fully-connected feedforward network consisting of two linear transformations, and a ReLU is arranged between the two linear transformations, namely
FFN(x)=max(0,xW1+b1)W2+b2 (2)
Where the symbol w denotes the weight matrix, b denotes the basis vectors and the subscripts denote the different layers of the fusion.
The specific process of the step 4 is as follows: inputting the denoised string feature map and the enhanced and fused fault feature map into a cross feature enhancement module CFA, repeating the fusion part for 4 times under the action mechanism of the CFA, and outputting the fused feature map;
among them, the CFA mechanism of action is summarized as:
wherein, XqIs a vector representation form, P, of the enhanced fused fault characteristic diagramqIs corresponding to XqSpatial position coding of, XkvIs a denoised string feature map vector representation form, pkvIs XkvThe spatial encoding of the coordinates is performed,is output by fusion of multi-headed intersection features, XCFIs the output result of the CFA. According to the above formula, CFA is according to XkvAnd XqCalculates an attention map from a plurality of scales in between, and then calculates X from the attention mapkvReweighed and added to XqIn order to enhance the representation capability of the feature map.
The specific process of the step 5 is as follows: inputting the fused feature map into a prediction head, classifying through binary cross entropy loss in a classification branch, and performing regression training in a regression branch to obtain a fault foreground/background classification result;
wherein the classification branch is defined as:
Lcls=-∑[yilog(pj)+(1-yi)log(1-pj)] (4)
yjgroup-truth tag, y, representing the jth sample j1 denotes foreground, pjRepresenting the probability of belonging to the foreground predicted by the learning model;
the regression loss in the regression branch is defined as:
And 6, outputting the fault type and position of the photovoltaic string according to the thermodynamic diagram and the classification result. The specific process is as follows: obtaining a photovoltaic string prediction position according to the thermodynamic diagram, and tracking the serial number of the string according to the photovoltaic string prediction position; and simultaneously, obtaining a fault foreground/background classification result according to the step 5, and outputting a classification and tracking result of the faults in the string.
Examples
Aiming at the photovoltaic power generation industry in the new energy field, verification is carried out on a distributed photovoltaic system (a factory building roof or a resident roof) by combining with unmanned aerial vehicle shooting, by tracking 10 groups of 33 m-height photovoltaic string videos by using the method, each video is about 255 frames, the acquired original image of the photovoltaic string is shown in figure 5, a white area selected by a serial number frame is a string, a small square in the string is called a panel, the faults are generally divided into string faults (the whole string is a fault), panel faults, hot plate faults (bright spots appearing in the panel) and stripe faults (bright stripes in the panel), the thermodynamic diagram of the string is obtained by extracting and fusing the characteristics of the string and combining a generalized GNET network, and the result prediction of the group string is completed according to the thermodynamic diagram, the effect is shown in the part of the attached figure 6, and the tracking number information of the group string can be obtained according to the figure 6. Obtaining a fusion vector of internal fault information of the string by performing feature extraction, context feature fusion and string and fault feature fusion crossed by background information on fault faults, finishing identification and positioning of the fault information by combining classification and regression in the step 5, outputting internal fault position information of the string (comprising corresponding string numbers, fault positions (x _ starting point horizontal coordinate, y _ vertical coordinate, w _ fault width and h _ fault height)), finishing identification of the faults, and marking on a frame, wherein the effect is shown in figure 7 of the attached drawing part, so that tracking of the string numbers and tracking of the faults inside the string are finished, and the accuracy is not lower than 95%; through the data, the method provided by the invention is verified, on one hand, the management of string grouping can be facilitated, on the other hand, more accurate information about faults can be obtained, and the photovoltaic maintenance is facilitated.
Through the mode, the photovoltaic string tracking and fault tracking method based on the attention mechanism completes serial number tracking of the photovoltaic string and fault tracking output of the inside of the string by combining the characteristics of the attention mechanism and combining the characteristics of the pre-trained CNN tracking network based on the attention mechanism, can accurately describe the string and the fault information of the inside of the string, improves the detection accuracy, reduces the operation and maintenance cost, and facilitates the maintenance of workers.
Claims (8)
1. The attention mechanism-based photovoltaic string tracking and fault tracking method is characterized by comprising the following steps:
step 1, taking a plurality of groups of photovoltaic string pictures as samples, and training a CNN network tracking model based on an attention mechanism;
step 2, obtaining a target video, inputting the target video into a trained attention-based CNN network tracking model to obtain a video frame sequence, extracting the characteristics of the video frame sequence, and outputting a string characteristic diagram and a fault characteristic diagram of the video frame sequence;
step 3, combining an attention mechanism, performing feature enhancement on the string feature diagram, removing noise interference, performing enhancement fusion on the fault feature diagram, and inputting the denoised string feature diagram into a GNet network to obtain a thermodynamic diagram;
step 4, inputting the denoised string feature map and the enhanced and fused fault feature map into a cross feature enhancement module CFA, carrying out cross fusion based on attention, and outputting a fused feature map;
step 5, inputting the fused feature map into a prediction head, and classifying and regressing through the prediction head to obtain a fault foreground/background classification result;
and 6, outputting the fault type and position of the photovoltaic string according to the thermodynamic diagram and the classification result.
2. The attention mechanism-based photovoltaic string tracking and fault tracking method according to claim 1, wherein the specific process of the step 2 is as follows:
the method comprises the steps that a video of a photovoltaic string is collected by an unmanned aerial vehicle to serve as a target video, the target video is input into a trained CNN network tracking model based on an attention mechanism, string features are extracted through a high-dimensional feature extraction channel con-5 in the CNN network based on the attention mechanism to obtain a string feature map, and target image features are selected through a low-dimensional channel con-4 to extract internal fault detail information of the string to obtain a fault feature map.
3. The attention mechanism-based photovoltaic string tracking and fault tracking method according to claim 1, wherein the specific process of performing feature enhancement on the string feature map and performing enhancement fusion on the fault feature map in step 3 is as follows: combining an attention mechanism, performing feature enhancement on a string feature map or a fault feature map by using two context enhancement modules ECA, wherein the ECA is used for performing multi-head self-attention in a residual form, adaptively integrating information of different positions in the string feature map or the fault feature map, taking the string feature map or the fault feature map as input of the ECA mechanism, obtaining a denoised string feature map or an enhanced and fused fault feature map, and the action mechanism of the ECA expresses that:
XEC=X+MultiHead(X+Px,X+Px,X) (1)
wherein X represents the input of ECA, pxFor spatial position coding, XECIs the output of the ECA.
4. The attention-based photovoltaic string tracking and fault tracking method according to claim 1, wherein spatial position coding is used in the cross-feature enhancement module CFA in step 4, and an FFN module is introduced in the cross-feature enhancement module CFA, wherein the FFN module is a fully connected feedforward network consisting of two linear transforms with a ReLU in between, i.e. a ReLU between the two linear transforms
FFN(x)=max(0,xW1+b1)W2+b2 (2)
Where the symbol w denotes the weight matrix, b denotes the basis vectors and the subscripts denote the different layers of the fusion.
5. The attention mechanism-based photovoltaic string tracking and fault tracking method according to claim 1, wherein the specific process of step 4 is as follows: inputting the denoised string feature map and the enhanced and fused fault feature map into a cross feature enhancement module CFA, repeating the fusion part for N times under the action mechanism of the CFA, and outputting the fused feature map;
among them, the CFA mechanism of action is summarized as:
wherein, XqIs a vector representation form, P, of the enhanced fused fault characteristic diagramqIs corresponding to XqSpatial position coding of, XkvIs a denoised string feature map vector representation form, pkvIs XkvThe spatial encoding of the coordinates is performed,is output by fusion of multi-headed intersection features, XCFIs the output result of the CFA.
6. The attention-based photovoltaic string tracking and fault tracking method according to claim 1, wherein the prediction head in step 5 comprises a classification branch and a regression branch, each of which comprises a three-layer sensor having a hidden dimension d and a ReLU activation function.
7. The attention mechanism-based photovoltaic string tracking and fault tracking method according to claim 6, wherein the specific process of step 5 is as follows: inputting the fused feature map into a prediction head, classifying through binary cross entropy loss in a classification branch, and performing regression training in a regression branch to obtain a fault foreground/background classification result;
wherein the classification branch is defined as:
Lcls=-∑[yilog(pj)+(1-yi)log(1-pj)] (4)
yjgroup-truth tag, y, representing the jth samplej1 denotes foreground, pjRepresenting the probability of belonging to the foreground predicted by the learning model;
the regression loss in the regression branch is defined as:
8. The attention mechanism-based photovoltaic string tracking and fault tracking method according to claim 1, wherein the specific process of step 6 is as follows: obtaining a photovoltaic string prediction position according to the thermodynamic diagram, and tracking the serial number of the string according to the photovoltaic string prediction position; and simultaneously, obtaining a fault foreground/background classification result according to the step 5, and outputting a classification and tracking result of the faults in the string.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110687502.6A CN113496210B (en) | 2021-06-21 | 2021-06-21 | Photovoltaic string tracking and fault tracking method based on attention mechanism |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110687502.6A CN113496210B (en) | 2021-06-21 | 2021-06-21 | Photovoltaic string tracking and fault tracking method based on attention mechanism |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113496210A true CN113496210A (en) | 2021-10-12 |
CN113496210B CN113496210B (en) | 2024-02-02 |
Family
ID=77997343
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110687502.6A Active CN113496210B (en) | 2021-06-21 | 2021-06-21 | Photovoltaic string tracking and fault tracking method based on attention mechanism |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113496210B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117520924A (en) * | 2023-12-29 | 2024-02-06 | 国网浙江省电力有限公司舟山供电公司 | Island photovoltaic operation and maintenance fault cause analysis method and system based on multi-mode data |
CN117764223A (en) * | 2023-11-28 | 2024-03-26 | 北京潞电电力建设有限公司 | Photovoltaic operation method and system |
CN117764223B (en) * | 2023-11-28 | 2024-08-02 | 北京潞电电力建设有限公司 | Photovoltaic operation method and system |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20190228266A1 (en) * | 2018-01-22 | 2019-07-25 | Qualcomm Incorporated | Failure detection for a neural network object tracker |
CN111192292A (en) * | 2019-12-27 | 2020-05-22 | 深圳大学 | Target tracking method based on attention mechanism and twin network and related equipment |
CN112560656A (en) * | 2020-12-11 | 2021-03-26 | 成都东方天呈智能科技有限公司 | Pedestrian multi-target tracking method combining attention machine system and end-to-end training |
CN112560695A (en) * | 2020-12-17 | 2021-03-26 | 中国海洋大学 | Underwater target tracking method, system, storage medium, equipment, terminal and application |
-
2021
- 2021-06-21 CN CN202110687502.6A patent/CN113496210B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20190228266A1 (en) * | 2018-01-22 | 2019-07-25 | Qualcomm Incorporated | Failure detection for a neural network object tracker |
CN111192292A (en) * | 2019-12-27 | 2020-05-22 | 深圳大学 | Target tracking method based on attention mechanism and twin network and related equipment |
CN112560656A (en) * | 2020-12-11 | 2021-03-26 | 成都东方天呈智能科技有限公司 | Pedestrian multi-target tracking method combining attention machine system and end-to-end training |
CN112560695A (en) * | 2020-12-17 | 2021-03-26 | 中国海洋大学 | Underwater target tracking method, system, storage medium, equipment, terminal and application |
Non-Patent Citations (2)
Title |
---|
周双双;宋慧慧;张开华;樊佳庆;: "基于增强语义与多注意力机制学习的深度相关跟踪", 计算机工程, no. 02 * |
齐天卉;张辉;李嘉锋;卓力;: "基于多注意力图的孪生网络视觉目标跟踪", 信号处理, no. 09 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117764223A (en) * | 2023-11-28 | 2024-03-26 | 北京潞电电力建设有限公司 | Photovoltaic operation method and system |
CN117764223B (en) * | 2023-11-28 | 2024-08-02 | 北京潞电电力建设有限公司 | Photovoltaic operation method and system |
CN117520924A (en) * | 2023-12-29 | 2024-02-06 | 国网浙江省电力有限公司舟山供电公司 | Island photovoltaic operation and maintenance fault cause analysis method and system based on multi-mode data |
CN117520924B (en) * | 2023-12-29 | 2024-04-12 | 国网浙江省电力有限公司舟山供电公司 | Island photovoltaic operation and maintenance fault cause analysis method and system based on multi-mode data |
Also Published As
Publication number | Publication date |
---|---|
CN113496210B (en) | 2024-02-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111696075A (en) | Intelligent fan blade defect detection method based on double-spectrum image | |
CN113076816B (en) | Solar photovoltaic module hot spot identification method based on infrared and visible light images | |
CN115205264A (en) | High-resolution remote sensing ship detection method based on improved YOLOv4 | |
CN114092832A (en) | High-resolution remote sensing image classification method based on parallel hybrid convolutional network | |
CN110390308B (en) | Video behavior identification method based on space-time confrontation generation network | |
Han et al. | Detection of faults in solar panels using deep learning | |
CN115861619A (en) | Airborne LiDAR (light detection and ranging) urban point cloud semantic segmentation method and system of recursive residual double-attention kernel point convolution network | |
CN113283409A (en) | Airplane detection method in aerial image based on EfficientDet and Transformer | |
CN115131747A (en) | Knowledge distillation-based power transmission channel engineering vehicle target detection method and system | |
CN112700476A (en) | Infrared ship video tracking method based on convolutional neural network | |
CN113496210B (en) | Photovoltaic string tracking and fault tracking method based on attention mechanism | |
CN114529821A (en) | Offshore wind power safety monitoring and early warning method based on machine vision | |
CN112668662B (en) | Outdoor mountain forest environment target detection method based on improved YOLOv3 network | |
Su et al. | Local fusion attention network for semantic segmentation of building facade point clouds | |
CN114329031A (en) | Fine-grained bird image retrieval method based on graph neural network and deep hash | |
CN111539950A (en) | Solar panel defect detection system based on machine vision | |
CN116664549A (en) | Photovoltaic power station hot spot defect detection method based on feature perception | |
Song et al. | PDD: Post-Disaster Dataset for Human Detection and Performance Evaluation | |
CN110826478A (en) | Aerial photography illegal building identification method based on countermeasure network | |
CN114596273B (en) | Intelligent detection method for multiple defects of ceramic substrate by using YOLOV4 network | |
CN113284103B (en) | Substation equipment defect online detection method based on space transformation fast R-CNN model | |
Li et al. | Few-shot meta-learning on point cloud for semantic segmentation | |
Kan et al. | Real-Time domestic garbage detection method based on improved YOLOv5 | |
CN111353509B (en) | Key point extractor generation method of visual SLAM system | |
Xu et al. | A lightweight network of near cotton‐coloured impurity detection method in raw cotton based on weighted feature fusion |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |