Research and Application of an Algorithm for Identifying Hazards in UAV Inspection Images of High-Voltage Cable Channels

The underground high-voltage cable is a significant trend in the future development of urban areas, making the identification of hazards along the cable channels a critical research topic. With the rapid advancement of Unmanned Aerial Vehicle (UAV) technology and deep learning techniques, new methods for identifying hazards in cable channels have emerged. In this paper, novel solutions for two key tasks were proposed: identifying external mechanical damage and tree obstacles on high-voltage cable channels using computer vision technology. By collecting image data via UAVs, a dataset based on real-world environments was constructed. The tasks of external mechanical damage identification and tree obstacle recognition were accomplished using trained You-Only-Look-Once (YOLO) object detection and instance segmentation models. To select the most suitable computer vision model, the test results of YOLOv5 and YOLOv8 algorithms were evaluated in this paper, providing a comprehensive assessment of the two models in terms of accuracy, model size, and detection speed.