Instance Segmentation Method Based on YOLOv8 for Detecting Bird's Nest on Transmission Tower

With the widespread adoption of intelligent inspection technologies in transmission line maintenance, efficient detection of abnormal targets on transmission towers is critical for ensuring power grid safety. Bird nests on high-voltage towers pose severe risks including short circuits and equipment damage [1]. Traditional manual inspections suffer from low efficiency and poor real-time performance. This paper proposes an instance segmentation method based on YOLOv8 to address key challenges in detecting small-scale nests within complex backgrounds using UAV -captured high-resolution images. Experiments demonstrate significant improvements in both accuracy and recall compared to conventional approaches, providing an efficient technical pathway for automated transmission line inspection.