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
  • [21] GOU Junnian, DU Susu, WANG Shiduo, Et al., Lightweight feature fusion CenterNet transmission line insulator self-explosion detection, Journal of Beijing University of Aeronautics and Astronautics, pp. 1-13, (2022)
  • [22] HAN K, WANG Y H, TIAN Q, Et al., GhostNet: more features from cheap operations, Proceedings of 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 1580-1589, (2020)
  • [23] DUAN K W, BAI S, XIE L X, Et al., CenterNet: keypoint triplets for object detection, Proceedings of 2019 IEEE/CVF International Conference on Computer Vision, pp. 6569-6578, (2019)
  • [24] ZHAO Dahe, YAO Xiaotong, Identification of catenary insulator under complex background based on lightweight convolutional network, Insulators and Surge Arresters, 3, pp. 172-178, (2022)
  • [25] HOWARD A G, ZHU M, CHEN B, Et al., Mobilenets: Efficient convolutional neural networks for mobile vision applications, (2017)
  • [26] LIU W, ANGUELOV D, ERHAN D, Et al., SSD: single shot MultiBox detector, Proceedings of the 14th European Conference, pp. 21-37, (2016)
  • [27] IANDOLA F N, HAN S, MOSKEWICZ M W, Et al., SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and< 0.5 MB model size, (2016)
  • [28] ZHU X K, LYU S C, WANG X, Et al., TPH-YOLOv5: improved YOLOv5 based on transformer prediction head for object detection on drone-captured scenarios, Proceedings of 2021 IEEE/CVF International Conference on Computer Vision Workshops, pp. 2778-2788, (2021)
  • [29] CHEN J, KAO S, HE H, Et al., Run, Don’t walk: chasing higher FLOPS for faster neural networks, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 12021-12031, (2023)
  • [30] TANG Y, HAN K, GUO J, Et al., GhostNetv2: enhance cheap operation with long-range attention, Advances in Neural Information Processing Systems, 35, pp. 9969-9982, (2022)
1234