An ESRGAN-based Nest Detection Method for Transmission Lines

Transmission lines are the foundation of transmission networks, and damage to transmission lines can cause serious safety accidents. In recent years, the issue of bird's nests impacting transmission lines has been increasingly prominent. It has become crucial to detect and promptly remove bird's nests, making it a significant area of research in power inspection. However, the unique characteristics of bird's nests on transmission lines make it challenging to collect a substantial number of samples for building a deep learning object detection model using the existing power equipment dataset. Consequently, this limitation impedes the progress of relevant scientific research in this domain. In order to deal with the above problems, this paper proposes an ESRGAN-based transmission line nest detection method, which expands the transmission line nest samples by using the ESRGAN model, and improves the object detection algorithm combined with YOLOv5. As a result, the detection accuracy of transmission line nests increases from 76.8% to 82.5% on the dataset collected in this study, affirming the validity of the proposed method for the transmission lines nest detection.