现代电子技术2024，Vol.47Issue(2)：31-36,6.DOI:10.16652/j.issn.1004-373x.2024.02.007
基于改进YOLOv5的铁路接触网绝缘子检测方法
Method of railway catenary insulator detection based on improved YOLOv5
摘要
Abstract
Accurate identification of insulators in railway catenary is a crucial prerequisite for insulator defect detection.In order to address the problems of texture feature differences in different directions and uneven brightness in nighttime insulator images collected by the railway 4C system,a railway catenary insulator detection algorithm based on improved YOLOv5 is proposed.By means of the concept of cyclic exposure generation,the problem of uneven brightness on the image surface is solved,and the SREG module is designed to improve the uneven brightness on image surfaces.The C3 module in the backbone network is redesigned and incorporated rotation-invariant convolution to better extract the texture features of insulators in different directions.To validate the performance of the improved model,the experiments were conducted on a test set.The railway catenary insulator detection algorithm based on improved YOLOv5 can be applied to the identification of insulators with different texture directions,with an average recognition accuracy of 99.3%and an F1 value of 98.9%.It can effectively detect railway catenary insulators under the 4C system at night.关键词
铁路接触网绝缘子/目标检测/改进YOLOv5/SREG模块/C3模块/纹理特征提取Key words
railway catenary insulator/object detection/improved YOLOv5/SREG module/C3 module/texture feature extraction分类
信息技术与安全科学引用本文复制引用
聂晶鑫..基于改进YOLOv5的铁路接触网绝缘子检测方法[J].现代电子技术,2024,47(2):31-36,6.基金项目
中国铁建重大专项:高速铁路接触网系统智能监测技术研究(18-A02) （18-A02）
中铁第一勘察设计院科研课题:接触网无人机辅助巡检方案研究(17-35) （17-35）
