A lightweight enhanced YOLO framework for accurate small-target cable defect detection

The reliable detection of cable defects is critical for the safety and stability of power transmission systems. Traditional methods often struggle to identify small or blurred defects efficiently. This study presents YOLO-ESBD, an enhanced YOLO11-based framework for cable defect detection. The model integrates an optimized feature fusion strategy, an effective attention mechanism, and a lightweight detection head, improving the representation of small targets while maintaining real-time performance. Experiments on a specialized cable defect dataset show that YOLO-ESBD achieves a precision of 92.3%, recall of 92.1%, and mAP50 of 94.2%, with a model size of 5.9 MB and computational cost of 9.2 GFLOPs. These results indicate superior accuracy, robustness, and efficiency compared to mainstream algorithms. YOLO-ESBD demonstrates strong potential for practical deployment in automated power inspection, supporting intelligent monitoring and preventive maintenance of cable networks.