FloodWise: Autonomous Geotagging for Disaster Response

VISIMO proposes a feasibility study for the development of FloodWise, an AI decision-support tool that will enable professionals to effectively prepare, adapt, and respond to flooding events. FloodWise will autonomously scrape and geotag social media images of flooding events to produce flood depth estimation maps in real time. VISIMO will create a custom algorithm that will geotag social media images and videos in the absence of Exif data. Unlike current models, VISIMO’s model will pinpoint image locations with latitude/longitude coordinates, accurate to 20 meters, within narrow regions. VISIMO will seek to enhance and refine the custom flood depth estimation algorithm utilized in the BluPix flood response app. The BluPix model is not fully automated, and only estimates water depth using stop signs; VISIMO will expand this algorithm to apply to additional objects and process images containing reflections accurately. VISIMO’s study will determine (1) the ability to autonomously geotag social media data; (2) the expansion and refinement of the BluPix flood depth estimation algorithm; and (3) the overall feasibility of the proposed final architecture. VISIMO will produce a demonstration that will show, for five locations, the pipeline’s ability to accurately geotag social media images and videos as new observations for actionable flood intelligence.