Most people interact with maps regularly, such as when getting from point A to point B, tracking the weather, planning trips, etc. But beyond these everyday activities, maps are increasingly being combined with artificial intelligence to create powerful tools for urban modeling, navigation systems, predicting and responding to natural disasters, monitoring climate change, virtual habitat modeling, and other types of surveillance.
“Maps are an essential part of our lives,” says Aayush Dhakal, a graduate student at the McKelvey School of Engineering at Washington University in St. Louis, “They allow us to learn patterns and understand distributions across geographic space.”
Dhakal and Srikumar Sastry, also a graduate student in the McKelvey School of Engineering, are collaborating with Nathan Jacobs, professor of computer science and engineering, to develop models that use satellite imagery to support these efforts. Dhakal's project, Sat2Cap, allows users to create maps from free-form text descriptions. Sastry developed GeoSynth, a model that synthesizes satellite imagery based on a given text prompt or geographic location.
Dhakal and Sastry presented their findings at the EarthVision workshop held in conjunction with the Computer Vision and Pattern Recognition 2024 conference in Seattle on June 17 this year. EarthVision aims to advance machine learning-based analysis of remote sensing data, with a special focus on pressing challenges and applications, including monitoring natural disasters, urban growth, deforestation and climate change.
For more information, visit the McKelvey Engineering website.
