Rapid Detection Method for Self-exploding Defects in Glass Insulators Based on Improved FasterNet and YOLOv5

被引：0
作者：
Wu K. [1 ]
Xu Z. [1 ]
Shan H. [1 ]
机构：
[1] School of Electronics and Information Engineering, Lanzhou Jiaotong University, Lanzhou
来源：
Gaodianya Jishu/High Voltage Engineering | 2024年 / 50卷 / 05期
关键词：
BiFPN-F; defect detection; DFC Attention; FasterNet; PConv; YOLOv5s;
D O I：
10.13336/j.1003-6520.hve.20231022
中图分类号：
学科分类号：
摘要：
In order to realize the real-time and fast inspection of insulator defects in power transmission lines, a fast defect detection arithmetic FasterNet-YOLOv5 is proposed by combining FasterNet-tiny and YOLOv5-s-v6.1 network model improvement. Firstly, a FasterNet network with a small number of parameters and faster reasoning speed is introduced to replace the original CSPDarkNet53 backbone network to speed up the detection speed of the network. Then, the DFC-FasterNet module is designed in the backbone feature extraction network by combining the decoupled fully connected (DFC) mechanism proposed by the GhostNetv2 network, and the DFC attention mechanism in the module can increase the receptive field during the feature extraction process to improve the detection accuracy of the network. Finally, for the case of glass insulator self-blast defects with smaller targets and more complex background, the Neck module is redesigned, and the BiFPN-F feature fusion module is proposed to enable the network to more accurately localize the insulator defect region. The experimental results show that the improved algorithm can locate quickly and accurately, its mean average precision (mAP) reaches 93.3%, which is improved by 5.67% compared with the pre-improvement, and the detection speed reaches 45.7 Hz, which is nearly one times higher than the pre-improvement. Meanwhile, compared with the latest YOLOv8n and YOLOv7-tiny, the improved FasterNet-YOLOv5 has more advantages in detecting the self-destructive defects in terms of accuracy and speed, and the proposed algorithm can locate and identify insulators and their self-destructive defects in real time more quickly. © 2024 Science Press. All rights reserved.
引用
收藏
页码：1865 / 1876
页数：11
相关论文
共 38 条
  • [1] SHAO Guiwei, LIU Zhuang, FU Jing, Et al., Research progress in unmanned aerial vehicle inspection technology on overhead transmission lines, High Voltage Engineering, 46, 1, pp. 14-22, (2020)
  • [2] JIA Xiaofen, YU Yeqi, GUO Yongcun, Et al., Lightweight detection method of self-explosion defect of aerial photo insulator, High Voltage Engineering, 49, 1, pp. 294-300, (2023)
  • [3] WANG Yuqing, YUAN Tian, NIE Lin, Et al., Machine vision inspection method for defects of glass insulator, High Voltage Engineering, 48, 12, pp. 4933-4940, (2022)
  • [4] WU Kaijun, DING Yuan, Wide self-attention fusion of dense residual network for image defogging, Journal of Hunan University (Natural Sciences), 50, 8, pp. 13-22, (2023)
  • [5] GOU Junnian, DU Susu, LIU Li, Transmission line insulator self-explosion detection based on improved mask region-convolutional neural network, Transactions of China Electrotechnical Society, 38, 1, pp. 47-59, (2023)
  • [6] GE Z, LIU S T, WANG F, Et al., Yolox: exceeding yolo series in 2021, (2021)
  • [7] WANG C Y, BOCHKOVSKIY A, LIAO H Y M., YOLOv7: trainable bag-of-freebies sets new state-of-the-art for real-time object detectors, Proceedings of 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 7464-7475, (2023)
  • [8] GIRSHICK R., Fast R-CNN, Proceedings of 2015 IEEE International Conference on Computer Vision, pp. 1440-1448, (2015)
  • [9] MA Xuesen, MA Ji, JIANG Gonghui, Et al., Insulator defect detection method based on attention mechanism and multi-scale feature fusion, Journal of Nanjing University (Natural Science), 58, 6, pp. 1020-1029, (2022)
  • [10] GU R, WANG G T, SONG T, Et al., CA-Net: comprehensive attention convolutional neural networks for explainable medical image segmentation, IEEE Transactions on Medical Imaging, 40, 2, pp. 699-711, (2021)
1234