A Lightweight Algorithm for Insulator Target Detection and Defect Identification
The accuracy of insulators and their defect identification by UAVs (unmanned aerial vehicles) in transmission-line inspection needs to be further improved, and the model size of the detection algorithm is significantly reduced to make it more suitable for edge-end deployment. In this paper, the algo...
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-01-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/23/3/1216 |
| _version_ | 1797623235874914304 |
|---|---|
| author | Gujing Han Liu Zhao Qiang Li Saidian Li Ruijie Wang Qiwei Yuan Min He Shiqi Yang Liang Qin |
| author_facet | Gujing Han Liu Zhao Qiang Li Saidian Li Ruijie Wang Qiwei Yuan Min He Shiqi Yang Liang Qin |
| author_sort | Gujing Han |
| collection | DOAJ |
| description | The accuracy of insulators and their defect identification by UAVs (unmanned aerial vehicles) in transmission-line inspection needs to be further improved, and the model size of the detection algorithm is significantly reduced to make it more suitable for edge-end deployment. In this paper, the algorithm uses a lightweight GhostNet module to reconstruct the backbone feature extraction network of the YOLOv4 model and employs depthwise separable convolution in the feature fusion layer. The model is lighter on the premise of ensuring the effect of image information extraction. Meanwhile, the ECA-Net channel attention mechanism is embedded into the feature extraction layer and PANet (Path Aggregation Network) to improve the recognition accuracy of the model for small targets. The experimental results show that the size of the improved model is reduced from 244 MB to 42 MB, which is only 17.3% of the original model. At the same time, the mAp of the improved model is 0.77% higher than that of the original model, reaching 95.4%. Moreover, the mAP compared with YOLOv5-s and YOLOX-s, respectively, is improved by 1.98% and 1.29%. Finally, the improved model is deployed into Jetson Xavier NX and run at a speed of 8.8 FPS, which is 4.3 FPS faster than the original model. |
| first_indexed | 2024-03-11T09:25:48Z |
| format | Article |
| id | doaj.art-05c070c9e5b5442cbf2f35a129b4ed86 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-11T09:25:48Z |
| publishDate | 2023-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-05c070c9e5b5442cbf2f35a129b4ed862023-11-16T17:57:49ZengMDPI AGSensors1424-82202023-01-01233121610.3390/s23031216A Lightweight Algorithm for Insulator Target Detection and Defect IdentificationGujing Han0Liu Zhao1Qiang Li2Saidian Li3Ruijie Wang4Qiwei Yuan5Min He6Shiqi Yang7Liang Qin8Department of Electronic and Electrical Engineering, Wuhan Textile University, Wuhan 430200, ChinaDepartment of Electronic and Electrical Engineering, Wuhan Textile University, Wuhan 430200, ChinaState Grid Information & Telecommunication Group Co., Ltd., Beijing 102211, ChinaDepartment of Electronic and Electrical Engineering, Wuhan Textile University, Wuhan 430200, ChinaDepartment of Electronic and Electrical Engineering, Wuhan Textile University, Wuhan 430200, ChinaDepartment of Electronic and Electrical Engineering, Wuhan Textile University, Wuhan 430200, ChinaSchool of Electrical and Automation, Wuhan University, Wuhan 430072, ChinaSchool of Electrical and Automation, Wuhan University, Wuhan 430072, ChinaSchool of Electrical and Automation, Wuhan University, Wuhan 430072, ChinaThe accuracy of insulators and their defect identification by UAVs (unmanned aerial vehicles) in transmission-line inspection needs to be further improved, and the model size of the detection algorithm is significantly reduced to make it more suitable for edge-end deployment. In this paper, the algorithm uses a lightweight GhostNet module to reconstruct the backbone feature extraction network of the YOLOv4 model and employs depthwise separable convolution in the feature fusion layer. The model is lighter on the premise of ensuring the effect of image information extraction. Meanwhile, the ECA-Net channel attention mechanism is embedded into the feature extraction layer and PANet (Path Aggregation Network) to improve the recognition accuracy of the model for small targets. The experimental results show that the size of the improved model is reduced from 244 MB to 42 MB, which is only 17.3% of the original model. At the same time, the mAp of the improved model is 0.77% higher than that of the original model, reaching 95.4%. Moreover, the mAP compared with YOLOv5-s and YOLOX-s, respectively, is improved by 1.98% and 1.29%. Finally, the improved model is deployed into Jetson Xavier NX and run at a speed of 8.8 FPS, which is 4.3 FPS faster than the original model.https://www.mdpi.com/1424-8220/23/3/1216lightweight algorithmYOLOv4GhostNetinsulatorattention mechanism |
| spellingShingle | Gujing Han Liu Zhao Qiang Li Saidian Li Ruijie Wang Qiwei Yuan Min He Shiqi Yang Liang Qin A Lightweight Algorithm for Insulator Target Detection and Defect Identification Sensors lightweight algorithm YOLOv4 GhostNet insulator attention mechanism |
| title | A Lightweight Algorithm for Insulator Target Detection and Defect Identification |
| title_full | A Lightweight Algorithm for Insulator Target Detection and Defect Identification |
| title_fullStr | A Lightweight Algorithm for Insulator Target Detection and Defect Identification |
| title_full_unstemmed | A Lightweight Algorithm for Insulator Target Detection and Defect Identification |
| title_short | A Lightweight Algorithm for Insulator Target Detection and Defect Identification |
| title_sort | lightweight algorithm for insulator target detection and defect identification |
| topic | lightweight algorithm YOLOv4 GhostNet insulator attention mechanism |
| url | https://www.mdpi.com/1424-8220/23/3/1216 |
| work_keys_str_mv | AT gujinghan alightweightalgorithmforinsulatortargetdetectionanddefectidentification AT liuzhao alightweightalgorithmforinsulatortargetdetectionanddefectidentification AT qiangli alightweightalgorithmforinsulatortargetdetectionanddefectidentification AT saidianli alightweightalgorithmforinsulatortargetdetectionanddefectidentification AT ruijiewang alightweightalgorithmforinsulatortargetdetectionanddefectidentification AT qiweiyuan alightweightalgorithmforinsulatortargetdetectionanddefectidentification AT minhe alightweightalgorithmforinsulatortargetdetectionanddefectidentification AT shiqiyang alightweightalgorithmforinsulatortargetdetectionanddefectidentification AT liangqin alightweightalgorithmforinsulatortargetdetectionanddefectidentification AT gujinghan lightweightalgorithmforinsulatortargetdetectionanddefectidentification AT liuzhao lightweightalgorithmforinsulatortargetdetectionanddefectidentification AT qiangli lightweightalgorithmforinsulatortargetdetectionanddefectidentification AT saidianli lightweightalgorithmforinsulatortargetdetectionanddefectidentification AT ruijiewang lightweightalgorithmforinsulatortargetdetectionanddefectidentification AT qiweiyuan lightweightalgorithmforinsulatortargetdetectionanddefectidentification AT minhe lightweightalgorithmforinsulatortargetdetectionanddefectidentification AT shiqiyang lightweightalgorithmforinsulatortargetdetectionanddefectidentification AT liangqin lightweightalgorithmforinsulatortargetdetectionanddefectidentification |