基于改进YOLOv7的输电线路绝缘子缺陷检测方法OA

As a key device in power systems,insulator defect detection plays a crucial role in ensuring the safe operation of power systems.Existing detection methods rely on traditional image processing technologies and deep learning models,but there are deficiencies in detection accuracy,speed,and robustness.To address the above is-sues,this study proposes an insulator defect detection method based on improved YOLOv7(You Only Look Once ver-sion 7).RepVGG(Re-parameterization Visual Geometry Group)is introduced into the backbone network of YOLOv7 to enhance feature extraction capability,and the DIoU(Distance Intersection over Union)loss function is a-dopted to optimize the regression accuracy of bounding boxes.Experimental results on the public insulator defect de-tection dataset show that the recall rate,mAP@0.5(mean Average Precision)and mAP@0.5:0.95 of the improved YOLOv7 model are increased by 5.8 percentage,0.9 percentage,and 3.4 percentage points respectively,which ver-ifies the effectiveness and superiority of the proposed method,and provides reliable technical support for the safe oper-ation of power systems.