Foreign Object Detection Algorithm for Transmission Lines Based on YOLO-World

Foreign object detection based on unmanned aerial vehicle (UAV) aerial images is an important part of intelligent inspection of transmission lines. The YOLO series of object detection algorithms have become mainstream methods due to their high accuracy and fast detection capabilities. However, they require predefined target categories, which limits their applicability in open scenarios. YOLO-World enhances its adaptability to open scenarios through visual-language modeling and large-scale dataset pre-training. However, the environment for foreign object detection in transmission lines is complex, and the types of targets are numerous, which are not fully covered in general datasets. To fully leverage the advantages of YOLOWorld, it must be fine-tuned to better learn and adapt to the specific foreign object dataset of transmission lines. Experimental results show that the fine-tuned YOLO-World significantly improves detection accuracy and reliability in actual inspections, effectively identifying various unknown foreign objects, thereby ensuring the safe operation of transmission lines.