Enhancing hyperspectral power transmission line defect and hazard identification with an improved YOLO-based model

• Published in: Optoelectronics letters Vol. 20; no. 11; pp. 681 - 688
• Main Authors: Wang, Meng, Sun, Long, Jiang, Jiong, Yang, Jinsong, Zhang, Xingru
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.11.2024; Springer Nature B.V; Ningbo Power Transmission and Transformation Construction Co.,Ltd.,Yongyao Technology Branch,Ningbo 315100,China%State Grid Ningbo Power Supply Company,Ningbo 315100,China
• Subjects: Algorithms; Deep learning; Defects; Dislocations; Hazard identification; Image enhancement; Lasers; Machine learning; Optical Devices; Optics; Photonics; Physics; Physics and Astronomy; Power lines; Real time; Remote monitoring; Remote sensing; Video transmission; A
• ISSN: 1673-1905; 1993-5013
• DOI: 10.1007/s11801-024-3213-3

To address the challenges of inefficient manual inspections and time-consuming video monitoring for power transmission lines, this paper presents an innovative solution. It combines deep learning algorithms with visible light remote sensing images to detect defects and hazards. Deep learning offers enhanced robustness, significantly improving efficiency and accuracy. The study utilizes you only look once version 7 (YOLOv7) as a foundational framework, enhancing it with the Transformer algorithm, Triplet Attention mechanism, and smooth intersection over union (SIoU) loss function. Experimental results show a remarkable 92.3% accuracy and an 18.4 ms inference speed. This approach promises to revolutionize power transmission line maintenance, offering real-time, high-precision defect and hazard identification.