A novel AI-powered flood damage assessment
Peer-Reviewed Publication
Updates every hour. Last Updated: 7-Nov-2025 20:11 ET (8-Nov-2025 01:11 GMT/UTC)
Researchers at The University of Osaka developed a deep learning model for rapid building damage assessment after floods using satellite imagery. This research establishes the first systematic benchmark for this task and introduces a novel semi-supervised learning method achieving 74% of fully supervised performance with just 10% of the labeled data. A new, lightweight deep learning model named Simple Prior Attention Disaster Assessment Net or SPADANet significantly reduces missed damaged buildings, improving recall by over 9% compared to existing models. This work provides crucial design principles for future AI disaster response, enabling faster and more efficient life-saving operations.
Researchers at The University of Texas at Austin analyzed the calcium isotopes in the teeth enamel of four different dinosaur species to discover what they ate. They found that some dinosaurs were discerning eaters, with different species preferring different plant parts. This helps explain how these dinosaurs, which all roamed the western U.S. during the Late Jurassic, were all able to coexist in the same ecosystem.
A treasure trove of exceptionally preserved early animals from more than half a billion years ago has been discovered in the Grand Canyon, one of the natural world’s most iconic sites. The rich fossil discovery – the first such find in the Grand Canyon – includes tiny rock-scraping molluscs, filter-feeding crustaceans, spiky-toothed worms, and even fragments of the food they likely ate.
Inspired by a hitchhiking fish that uses a specialized suction organ to latch onto other marine animals, MIT engineers designed a mechanical adhesive device that attaches to soft, slippery surfaces and remains there for days or weeks. The device could be used to deliver drugs in the GI tract or monitor aquatic environments.
Researchers at Institute of Industrial Science, The University of Tokyo, have taken a great stride in supporting earthquake prevention research by developing a system for seafloor position measurements with centimeter-level precision. Combining the Global Navigation Satellite System–Acoustic and an unmanned aerial vehicle, the proposed system eliminates the need for manned surface vessels.
This paper proposes a new theoretical content and research framework of multi-spheric interaction-driven hydrocarbon formation and enrichment through in-depth analyses of the Earth’s multi-spheric coupling mechanisms and cross-spheric cycling processes of volatiles. It establishes a novel theoretical paradigm for optimizing target prioritization of both mature field revitalization and frontier play assessment.