Research on Insulator Defect Detection in Power Inspection Images Based on PaddleDetection

Insulators primarily support transmission lines as a shielding control in the electric power system, which exposes them to the elements over an extended period of time and forces them to contend with their severe effects, making them prone to failure. When the insulator fails, it will significantly increase regular energy use and result in significant losses in output and life. In order to safeguard the regular and steady functioning of transmission lines, insulator defect issues must be identified early on during the power inspection process. This study uses insulator images captured during the inspection process and the PP-YOLOv2 deep learning algorithm to explore insulator defect detection in the context of UAV power inspection. The PP-YOLO algorithm is found to be more accurate than the YOLOv3 algorithm while guaranteeing real-time detection and can better extract the data features of insulators in the inspection images, which has better practicability, according to simulation tests on the data set.