Different degrees of pulverization will occur on the surface of silicone rubber composite insulators operating in harsh environments such as strong ultraviolet, high altitude, salt fog and high humidity, thus affecting the operation performance of insulators. In this paper, image of more than 2000 pulverized composite insulators operating in Zhejiang power grid is captured by UAV. The judgment principle of severe pulverization is determined according to the characteristics of pulverization image and the influence of pulverization on the operation performance of composite insulator. The identification method for composite insulator pulverization defect is established based on YOLOv5 algorithm, and the identification accuracy of severe pulverization reaches 96.53% through test samples.