12-05, 11:15–11:45 (America/Belem), Room I
Every geospatial project begins with a quest for answers. Large Language Models (LLMs) are revolutionizing how we can directly understand user needs through techniques like natural language to data structure conversion. Over the couple years, we have been exploring how AI could be used for working with geospatial data. What started as figuring out how to use natural language to make STAC queries to find public data has led to much more, including natural language geocoding to contextual image searching of public data such as NAIP and Sentinel-2.
In this talk we will explore how AI can be used to help automate some of geospatial’s most tedious tasks using open data, and how open vision models can be combined to create powerful tools for search & discovery of earth imagery.
This talk will include an overview of AI for use in geospatial analysis, with a focus on using open data and open models. We will show some live demos to create accurate AOIs with natural language, as well as for advanced searching of landscape features in public datasets. Additionally we will give an overview of techniques like Retrieval Augmented Generation and LLM Agents and the potential for how these may be used to transform geospatial data science.
Matt Hanson is the Director of Aerospace at Element 84, a commercial geospatial consultancy that utilizes open-source to build solutions. With an education in Remote Sensing at the Rochester Institute of Technology, he has been working with geospatial data for over 25 years. As an author and contributor to multiple open-source projects (starting with GeoNode in 2012), he has gone on to help create open standards, like STAC, as well as the open-source ecosystem around data interoperability.
A frequent speaker at geospatial conferences, this will be Matt's 10th international FOSS4G conference.