Transmission Line Insulator Defect Detection with Improved YOLOv11

To address the challenge of insufficient detection performance in identifying transmission line insulator defects under complex outdoor conditions, this study proposes an enhanced YOLOv11-based model. First, the Res2-C3k2 module is integrated into the YOLOv11 backbone, which strengthens the extraction of high-resolution feature representations and improves the model’s perception capability. Second, the traditional PAN structure is replaced by a redesigned BiFPN, which significantly enhances the effectiveness of multi-scale feature fusion and contributes to improved detection precision. Additionally, the original CIoU loss is substituted with the WIoU loss function, which introduces a dynamic weighting strategy to reduce the detrimental effect of low-quality anchor boxes on reliable predictions. Experimental evaluations on a self-curated dataset indicate that the proposed model achieves performance gains of 1.3% in precision, 1.3% in recall, 2.2% in mAP@0.5, and 4.4% in mAP@0.5:0.95, outperforming the baseline YOLOv11. Overall, the proposed framework demonstrates significant improvements in detection capability and offers practical potential for deployment in real-world insulator monitoring applications.