Detection of Damaged Insulator Based on Improved Cooratt-yolov5s

Abstract

Insulators are insulating materials used in the construction of electrical transmission systems. They play vital roles in high-voltage transmission lines. The degree of insulators’ damage is related to the stability of the whole power supply line. Therefore, regular inspection of insulators along transmission lines is necessary. The traditional manual inspection is costly and inefficient. There is a great prospect of replacing manual inspection by unmanned aerial vehicles inspection. To address the problems of complex backgrounds and low damage identification in insulator images as we limited arithmetic power of UAV, this paper proposes an improved Cooratt-yolov5s algorithm model to achieve the rapid detection of damaged insulators, which adds Cooratt-attention module to yolov5s backbone network to strengthen the recognition ability of small damage. In the experiment, compared with the original model, Cooratt-yolov5s model has a stable improvement in mAP index and detection speed, which can accomplish the task of real-time and accurate detection of insulator damage, and has a good reference significance for power companies to improve the traditional inspection methods.