Comparing Image Segmentation Models for Flood Inundation Mapping

Flood inundation mapping plays a pivotal role in disaster management and mitigation efforts. Accurate identification and segmentation of flood pixels are crucial for assessing the extent of inundation and aiding timely response efforts. In this study, we present a comparative analysis of various segmentation models for flood pixel segmentation, focusing on their efficacy in flood inundation mapping tasks. The models we compare are: U-Net (2 approaches), LinkNet, IBM-NASA’s Prithvi (2 approaches), and SegFormer.