Based on YOLOv5-Insulators for Insulator Anomaly Detection

Insulators, which are crucial components within power systems, need precise state detection so as to guarantee the secure and steady functioning of the grid. Traditional insulator detection methods face challenges such as insufficient accuracy in complex environments. This paper proposes a method named YOLOv5-Insulators for insulator detection, which enhances the original YOLOv5 architecture by incorporating the Swin-Transformer framework to improve target detection performance and introduces the MPDIoU loss function to enhance model robustness. Experimental results show that the YOLOv5-Insulators model can proficiently identify insulators with an accuracy rate of 95.6% and a mean average precision (mAP) of 92.5%, demonstrating excellent robustness and effectively improving the efficiency and reliability of insulator detection in power systems.