Insulator Defect Detection of Transmission Lines Based on Improved Fused YOLOv11s

The detection of insulators in transmission lines is critical for ensuring safe operation of power grids. This study proposes an improved YOLOv11s algorithm that effectively identifies two types of insulator defects (damage and flashover) with a high detection accuracy and rapid inference speed. The enhanced algorithm incorporates an inception depthwise convolution (IDC) module, embeds a multi‐scale efficient multiscale attention (EMA) mechanism into the backbone network, and introduces a dynamic feature‐aware hybrid loss (DFHL) function. The experimental results showed that the mean average precision (mAP@0.5) reached 0.988, and mAP@0.5:0.95 achieved 0.938, representing improvements of 5% and 7.4% compared to the original YOLOv11s model. Comparative analysis with other object detection algorithms demonstrates the superiority of the proposed method: its mAP@0.5 is 10.3% higher than that of the SSD algorithm, while maintaining lightweight characteristics (8.9 M parameters, 21.3G FLOPs) and real‐time performance (51 FPS). These results confirm that the improved algorithm excels in terms of both accuracy and speed for insulator‐defect detection. © 2026 Institute of Electrical Engineers of Japan and Wiley Periodicals LLC.