To address the challenges of complex backgrounds and uncertainties in target scales in insulator defect detection during unmanned aerial vehicle (UAV) inspection of transmission lines

this paper proposes a WMEF-YOLO-based defect image detection method. Firstly

a WTC3k2 module is designed to expand the model’s receptive field through multi-frequency feature fusion

enhancing the ability to capture irregular damage edges of insulators. Then

a C2PSA_M module is developed by integrating C2PSA with a multi-scale dilated attention mechanism

enhancing the feature extraction performance under complex backgrounds. Furthermore

an efficient up-convolution block (EUCB) is introduced to optimize target feature reconstruction

ensuring effective retention of fine-grained details during the decoding stage. Finally

an F-MPDIoU loss function

in which geometric constraints and dynamic sample weighting mechanism are incorporated

is proposed to effectively enhance the bounding box regression accuracy and training stability. Experimental results on a self-built insulator defect dataset demonstrate that the improved model outperforms the baseline in both accuracy and real-time performance

achieving 5.3% higher than mAP@0.5

3.7% higher than mAP@0.95

and an increase of 13 frames per second. The superior generalization capability has also been validated on multiple datasets. The study provides novel insights for intelligent UAV inspection systems for intelligent UAV inspection systems.