Article: Bidirectional attention network for real-time segmentation of forest fires based on UAV images Journal: International Journal of Information and Communication Technology (IJICT) 2024 Vol.25 No.6 pp.38 - 51 Abstract: For the purpose of monitoring hill fires, a bidirectional attention module is designed, which not only allocates the weights between different features reasonably, but also pays attention to the relationship between neighbouring pixels of the same feature to accurately segment the boundaries of flames and forest background. A feature fusion module is designed to effectively fuse deep and shallow feature information, and automatically adjust the input feature weights to improve the model's segmentation ability for small fire targets. Finally, we conducted experiments on the public dataset Flame, and the results show that the designed model outperforms other state-of-the-art methods in segmentation accuracy and computational efficiency. Inderscience Publishers - linking academia, business and industry through research

Title: Bidirectional attention network for real-time segmentation of forest fires based on UAV images

Authors: Zhuangwei Ji; Xincheng Zhong

Addresses: Computer Science Department, Changzhi College, Changzhi, 046000, Shanxi, China ' Computer Science Department, Changzhi College, Changzhi, 046000, Shanxi, China

Abstract: For the purpose of monitoring hill fires, a bidirectional attention module is designed, which not only allocates the weights between different features reasonably, but also pays attention to the relationship between neighbouring pixels of the same feature to accurately segment the boundaries of flames and forest background. A feature fusion module is designed to effectively fuse deep and shallow feature information, and automatically adjust the input feature weights to improve the model's segmentation ability for small fire targets. Finally, we conducted experiments on the public dataset Flame, and the results show that the designed model outperforms other state-of-the-art methods in segmentation accuracy and computational efficiency.

Keywords: UAV images; semantic segmentation; hill fire detection.

DOI: 10.1504/IJICT.2024.141434

International Journal of Information and Communication Technology, 2024 Vol.25 No.6, pp.38 - 51

Received: 02 Jul 2024
Accepted: 23 Jul 2024
Published online: 12 Sep 2024 *

Title: Bidirectional attention network for real-time segmentation of forest fires based on UAV images

Keep up-to-date