Michael Shensky
GIS & Geospatial Data Coordinator
University of Texas at Austin
Aaron Choate
Director of Research & Strategy
University of Texas at Austin
Danna Gurari
Assistant Professor
University of Colorado Boulder
In fall 2020 work began on the Machine Learning for Geographic Information System (ML4GIS) project, which was envisioned as a multiyear effort to create software, workflows, and datasets that would allow the University of Texas Libraries and its project partners to explore the development of machine learning algorithms for georeferencing and extracting information from scanned maps images to facilitate their use in GIS software. This project brief will describe the significant progress made during the first year of this effort, which has resulted in the successful development of a new open source application that can be utilized to generate annotation data for later use in developing machine learning models. Particular focus will be placed on the capabilities and potential of this new application which provides a streamlined, customizable interface that can be used locally or deployed in Amazon Mechanical Turk to enable scalable crowdsourced annotation development for large collections of scanned map images.