Advanced YOLOv10 for Aerial Power Line Fault Detection

As the dependency on electric energy by power systems increases, ensuring the safe and stable operation of aerial power lines becomes crucial. Traditional manual inspection methods are inefficient and costly. This paper proposes an aerial power line fault detection method based on an improved YOLOv10 algorithm, aiming to enhance the precision and efficiency of detection. The proposed method introduces an RCS-OSA module to enhance the network’s sensitivity to multi-scale targets, employs the SEAM Attention Module to compensate for information loss due to occlusions, and incorporates the Normalized Gaussian Wasserstein Distance to improve the identification of small objects. Experiments were conducted using a drone inspection image dataset provided by a power grid company. These tests demonstrated that the method achieved a 2.1% increase in key detection accuracy, measured by the mAP@50 metric, while maintaining high detection speed. This shows the method’s great potential for practical application.