YOLEF:一种面向输电线路红外图像的高精度实例分割方法OA

The security of transmission lines is related to the stability of the power grid and the continuity of power supply.Affected by long-term high-load operation and field environment,key components of the line are prone to overheating and other faults,and efficient and intelligent inspection methods are urgently needed to monitor them.Aiming at the problem that it is difficult to identify the key components of unmanned aerial vehicle(UAV)inspection on infrared images,this paper proposes a new instance segmentation model:you only look at electric fault(YOLEF).The model uses EfficientFormerV2 to replace the backbone network of you only look once(YOLO)11.At the same time,a lightweight joint feature dynamic selection head(JFDS-Head)detection head is designed by combining the feature dynamic guidance mechanism and the shared convolution structure.The experiment results show that:compared with the baseline model YOLO11n-seg,the overall segmentation accuracy of the proposed model on seven key components is increased by 4.0 percentage points,the recall rate is increased by 0.27 percentage points,the mAP@0.5 is increased by 3.74 percentage points,and the mAP@[0.5:0.95]is increased by 4.46 percentage points.It has the advantages of high accuracy and suitable for mobile deployment in infrared image segmentation tasks.