Power Line and Insulator Detection Using YOLOv5 with SE Attention Mechanism

This study aims to explore the importance and methods of detecting foreign object coverage on power lines and insulators, emphasizing their critical role in ensuring the safe and stable operation of electrical power systems. The research utilizes unmanned aerial vehicle (UAV) power inspections combined with deep learning technology, specifically the YOLOv5 algorithm integrated with the SE attention mechanism, to propose an efficient detection scheme for foreign objects and insulators on power lines. By adjusting the weights of channels in the convolutional feature maps, this method significantly enhances the model's ability to recognize small targets in complex backgrounds. Experimental results on a custom dataset demonstrate the superior performance of this approach, proving its practical value in power line inspection and maintenance. Moreover, by comparing the performance with other object detection models, the study further validates the distinct advantages of the YOLOv5 enhanced with the SE module across multiple evaluation metrics. These findings provide valuable references for power line inspection and maintenance, contributing to the technological advancement of electrical power system safety management.