Railway Track Defect Detection Based on YOLO11

With increasing train speeds and denser railway networks, intelligent detection of structural defects in railway tracks has become critical for transportation safety. To efficiently identify fine-grained defects such as cracks, missing bolts, and loose screws, this study proposes an automatic railway track defect detection method based on the lightweight object detection model YOLO11. A custom railway image dataset containing five representative defect categories was constructed and annotated, and the YOLO11 architecture was analyzed and optimized for deployment. Experimental results show that the proposed model achieves mAP@0.5 of 0.958, average precision of 0.957, and recall of 0.965, demonstrating excellent detection performance, especially for small objects. Ablation studies further explore the impact of model architecture, input resolution, and other factors, offering theoretical insights and practical guidance for optimization. The results confirm that YOLO11 exhibits strong robustness and adaptability, supporting its application in intelligent railway inspection.