Research on insulator detection algorithm of YOLOv5 fused with attention mechanism
Ming Zhang, Qingxing Li · 2022
In high-voltage transmission lines, the status of insulators needs to be monitored regularly to prevent fault, and the detection of insulators is more difficult under mist conditions. In this paper, an improved lightweight YOLOV5 algorithm is proposed to achieve defect detection of insulators. First of all, the MSR algorithm is incorporated in the YOLO model to achieve the anti-fogging effect, and in order to better detect the insulator, the CBAM attention module is introduced in the Backbone part of the YOLOv5 model. Experiments show that compared with the original model, the improved YOLOv5s-CBAM detection accuracy has been improved, and the mAP value has reached 98.4%, which can better complete the insulator detection task.