Insulator and Damage Detection and Location Based on YOLOv5

Abstract

In order to quickly and accurately detect and locate the insulator and its damage in the transmission line, by studying the YOLO (you only look once) series target detection algorithm model based on deep learning, a detection method based on YOLOv5 is proposed to achieve fast and accurate detection. By applying the network to the aerial insulator data set for training, the experimental results show that the highest AP (Average Precision) value based on YOLOv5 insulator detection is 96.47%, the highest AP value of insulator damage is 99.17%, and the overall m-AP (mean Average Precision) value is 97.82%. At the same time, YOLOv5s has a higher detection rate, and the real-time detection speed is 43.2FPS (Frames Per Second). The experimental results show that the target detection network based on YOLOv5 series has higher accuracy and faster calculation speed for transmission line insulator detection and damage identification under complex background, at the same time, the lightweight model of YOLOv5s is conducive to the deployment of UAV (Unmanned Aerial Vehicle) end model and improve inspection efficiency.