Bird's nest defect detection of transmission lines based on domain knowledge and occlusion reasoning

• DOI: 10.1049/gtd2.13007
• Publication Date: 2023-10-04T09:02:25Z
• Authors: Na Dong; Wenjing Zhang; Ze Chen; Haiyan Feng; Jiandong Jia
• Keywords: Feature (linguistics)

Abstract Bird's nest defect is an important cause of transmission line faults. To achieve accurate detection bird defects in complex scenarios, a model for lines was proposed that combines domain knowledge and occlusion reasoning networks. On the one hand, utilized location bird's nest, using edge to obtain tower area information constrain candidate frames. This helps reduce false caused by backgrounds. other on basis analyzing characteristics nests, employed networks randomly erase features at feature level simulate nests real scenes improve model's capability occluded targets. Additionally, multi‐scale fusion algorithm designed this paper adapt scale variations aerial images. Experimental results demonstrate outperforms advanced target models methods, with AP50 78.8% AR10 72.4% detection.