Lightweight YOLOv7 Real-Time Insulator Power Equipment Defect Detection Based on Attention Improvement

DOI

10.2991/978-94-6463-264-4_82How to use a DOI?

Keywords

Defect detection; insulator defects; small target detection; attention mechanism; power equipment

Abstract

efective faults in insulator power equipment can affect transmission equipment's normal operation and electricity consumption in the service area. To reduce or avoid transmission faults caused by defective insulator power equipment failures, structural defects of insula-tors need to be detected. In contrast, different insulator defects have significant differences in style and size. The defective parts in the UAV scenario have problems such as blurring, obscuration, and environmental factors, which lead to challenging insulator power equipment defect detection. Therefore, we propose a model specifically for detecting insulator defects in power equipment - GSNA-YOLOv7. We added GSConv to improve the New-Neck module to better balance the inference speed of the model with the detection accuracy of defective targets of power equipment, reduce the redundant information of the model, and better achieve the effect of real-time detection; improve the DownNAM module, introduce the attention mechanism, apply the weight sparsity penalty, stabilize the performance and computational efficiency, and make the model pay more attention to the defective small target information. The SFID insulator dataset and Visdrone2021 UAV dataset are trained and validated. The experimental results are analyzed, concluding that GSNA-YOLOv7 has a better detection effect for power equipment defect detection in the UAV shooting scenario and is more adaptable to detecting small targets in insulator fault defect datasets. The method is better than many existing insulator defect detections. The method outperforms numerous current insulator defect detection methods, with mAP improved by 0. 9% and 0.6% and parameter volume reduced by 0.2G, compared with the base model YOLOv7.

Copyright

© 2024 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Tao Wang
AU  - Jingfeng Xiao
AU  - Xinxin Meng
AU  - Wenzhong Yang
PY  - 2023
DA  - 2023/09/28
TI  - Lightweight YOLOv7 Real-Time Insulator Power Equipment Defect Detection Based on Attention Improvement
BT  - Proceedings of the 2023 3rd International Conference on Education, Information Management and Service Science (EIMSS 2023)
PB  - Atlantis Press
SP  - 710
EP  - 726
SN  - 2589-4900
UR  - https://doi.org/10.2991/978-94-6463-264-4_82
DO  - 10.2991/978-94-6463-264-4_82
ID  - Wang2023
ER  -