摘要
绝缘子是输电线路无人机巡检的重要目标。为了快速准确识别巡检图像中的绝缘子,通过采集巡检图像,构造绝缘子数据集,提出了一种基于深度学习单阶图像识别框架(YOLOV3-SPP)的绝缘子在线识别模型。该网络模型在YOLOV3的模型上加入了SPP模块。通过在Pytorch环境下的训练和测试,结果表明:加入SPP模块后提高了检测的精确度,并可以快速有效地完成绝缘子识别定位。
The insulator is an important target for UAV transmission line inspection.In order to quickly and accurately identify insulators in patrol images,an online insulator recognition model based on deep learning single order image recognition framework(YOLOV3-SPP)is proposed through collecting patrol images and constructing insulator data sets.This network model adds SPP module to the YOLOV3 model.Through the training and testing in Pyorch environment,the results show that after adding SPP module the detection accuracy can be improved,and the insulator identification and positioning can be completed quickly and effectively.
作者
唐睿
张铭予
徐宏
李微
李文波
TANG Rui;ZHANG Mingyu;XU Hong;LI Wei;LI Wenbo(Yan’an Power Supply Branch of Shaanxi Local Electric Power(Group)Co.,Ltd.,Yan’an 716000,Shaanxi,China;Xi’an University of Technology,Xi’an 710048,Shaanxi,China)
出处
《电网与清洁能源》
北大核心
2021年第4期41-46,共6页
Power System and Clean Energy
基金
国家自然科学基金项目(51779206)。
