期刊
ENERGY REPORTS
卷 8, 期 -, 页码 742-753出版社
ELSEVIER
DOI: 10.1016/j.egyr.2022.03.160
关键词
Feature pyramid; Visual saliency; Loss function; Automatic recognition of bird's nest; Image multi-channel fusion
This paper proposes a bird's nest recognition method that combines visual saliency and depth learning. The method achieves accurate identification and robustness, providing guidance for the operation and maintenance of transmission lines.
The automatic identification of bird's nest in the inspection image of transmission line is of great significance to the safe operation of transmission line. In this paper, a bird's nest recognition method which combines visual saliency and depth learning is proposed. This method not only has the advantage of rich feature information of visible light image, but also has the advantage of significant bird's nest target. The experimental results show that this method can accurately identify the images with different background, tower shape, shooting angle and shooting distance, and has good robustness and generalization, and the precision index values of Precision, Recall and IoU are 0.9622, 0.9465 and 0.9543 respectively. Compared with Faster R-CNN model, YOLO model and RetinaNet model, each index is greatly improved. This method is instructive to the operation and maintenance of transmission lines. (C) 2022 Published by Elsevier Ltd.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据