Transmission Line Defect Detection Based on Improved YOLO Neural Network

To address the frequent faults in transmission lines and enhance defect detection efficiency, we propose an improved YOLOv8 model combined with Inverted Residual Mobile Block (iRMB) and Wise-IOU. Compared with the traditional YOLOv8, this model significantly improves detection performance in complex scenarios by introducing iRMB and Wise-IOU modules. In YOLOv8, iRMB combines convolution operations, attention mechanisms, and skip connections to effectively enhance the detection of small-scale and complex background targets. This enables more accurate identification of cable defects and corroded towers in transmission lines, reducing missed and false detections. In addition to iRMB, the introduction of the Wise-IOU loss function further optimizes the target box regression process in YOLOv8. In transmission lines, when facing densely arranged cables and complex background interference, Wise-IOU can provide more precise bounding box adjustment strategies to ensure high-accuracy target detection. Experimental results show that the improved YOLOv8 with iRMB and Wise-IOU demonstrates excellent performance in transmission line defect detection, especially in complex scenarios such as insulator defects and faulty cables. This technology provides a more efficient and precise solution for the intelligent inspection of power grids.