A YOLOv8-based detection algorithm for insulator damage in transmission lines

In view of the complex background and small damage size of transmission line insulators in aerial images of unmanned aerial vehicles (UAVs), traditional detection algorithms usually expose such defects as false detection and missing detection. To solve this challenge, this research proposes an improved object detection algorithm based on YOLOv8, specially designed to detect insulator damage in transmission lines. Specifically, firstly, the Wise-IoU (SIoU) loss function is used to replace the Complete-IoU (CIoU) loss function in YOLOv8 to improve the positioning accuracy of the detection frame. Secondly, the small object detection layer is added to improve the recall rate of small defect objects. Finally, the large selective kernel network (LSK) module is introduced into the neck network of YOLOv8 to realize the dynamic adjustment of the spatial receptive field. The experimental results indicate that the improved model is 7%, 4.1%, and 1.6% higher than the original YOLOv8 in terms of precision, recall, and mean average precision (mAP). Compared with other mainstream object detection algorithms, the proposed model demonstrates a better detection effect. In other words, the proposed model can effectively support the real-time detection task of insulator status in transmission lines.