AUTHOR=Yadav Ritu , Nascetti  Andrea , Ban  Yifang 

TITLE=Deep attentive fusion network for flood detection on uni-temporal Sentinel-1 data

JOURNAL=Frontiers in Remote Sensing

VOLUME=Volume 3 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/remote-sensing/articles/10.3389/frsen.2022.1060144

DOI=10.3389/frsen.2022.1060144

ISSN=2673-6187

ABSTRACT=Floods are occurring across the globe and due to climate change, flood events are expected to increase in upcoming years. Current situations urge more focus on efficient monitoring of floods and detecting impacted areas.
In this study we propose two architectures for flood detection on uni-temporal Sentinel-1 SAR data. The first network presented in this work is 'Attentive U-Net' where feature maps are enhanced using spatial and channel-wise attention. This network works on VV, VH and the ratio of VV/VH. 'Attentive U-Net' yields 67% Intersection Over Union(IoU) on the Sen1Floods11 dataset, which is 3% better than the benchmark IoU.
The second proposed network is a dual-stream fusion network, where low-resolution elevation data, permanent water mask, and Sentinel-1 SAR (VV, VH) data are fused to extract flood areas. The 'Fusion network' improved IoU by 5% compared to the benchmark IoU. Our code is available on Github for reuse and experimentation.