Research on a Transmission Line Fault Detection Method Based on Improved YOLOv7-Tiny

In response to the challenges faced by most fault detection algorithms for transmission power lines, such as detecting small targets, issues of missed detections and false negatives, a large number of model parameters, and high computational requirements that hinder deployment on unmanned aerial vehicles, this study proposes a novel model called YOLOvS. YOLOvS addresses these challenges by incorporating lightweight models, introducing improved attention mechanisms, and utilizing an enhanced CIoU to achieve fast and accurate detection of typical faults in transmission power lines captured by aerial imagery. Experimental validation confirmed that the YOLOvS algorithm improves the detection accuracy of small targets while maintaining overall detection performance. Furthermore, it achieved a reduction of 21% in model size compared to YOLOv7-Tiny and a 20% decrease in computational requirements. These advancements significantly contribute to lowering the hardware costs associated with deploying the algorithm on unmanned aerial vehicles.