Researchers have introduced Urban Flood Observations (UFO), a new global dataset featuring hand-labeled post-flood inundation maps derived from high-resolution satellite imagery. The dataset contains 215 image chips from 14 flood events between 2017 and 2021, annotated for 'inundated' and 'non-inundated' areas. When used to train a segmentation model, UFO enabled an IoU of 77.3, significantly outperforming existing surface water products like NASA's IMPACT and Google's Dynamic World. AI
Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →
IMPACT Provides a benchmark dataset to improve AI-driven flood mapping accuracy and validation.
RANK_REASON This is a research paper describing a new dataset for flood inundation mapping.