The National Geospatial-Intelligence Agency has introduced a broad agency announcement to solicit innovative concepts that could help address GEOINT problems across the technical domains of foundational GEOINT, advanced phenomenologies and analytic technologies.
Under the Boosting Innovative GEOINT-Science and Technology BAA, NGA has issued a call for abstracts and proposals for the first topic: Geospatial-Intelligence Foundational Model — dubbed GFM — according to a solicitation notice published Friday.
The research topic seeks to build a trained foundational neural network model that can ingest overhead and ground-level imagery and vector data and help address geospatial questions while demonstrating a proof of concept.
Topic 1 has a one-year base period and three option years and covers GFM, multimodal machine learning, geospatial artificial intelligence, masked autoencoding, remote sensing, neural networks and geographic information system.
Questions for the GFM topic are due Jan. 2. NGA will accept abstracts until Jan. 23 and proposals through March 4 for the initial research topic and the performance period is expected to start on May 15.
