Uploaded December 2025 | Updated September 2026, 2 weeks ago
From Pixels to Earth Action: Operationalizing Geospatial AI Foundation Models
Universities Space Research Association
Our scientists will focus on recent efforts to operationalize geospatial AI foundation models for Earth Action, including the Prithvi-EO-2 foundation model which received the 2025 AGU Open Science Recognition Prize and Meta’s SAM2 foundation model for image segmentation that USRA is using in combination with remote sensing data for USGS use cases.
Prithvi’s development, led by NASA with collaborators from the geospatial AI community including USRA, focuses on high-resolution, multisensor satellite data. It is optimized for sustainability and climate applications—such as wildfire risk assessments, flood mapping, ecosystem monitoring, and climate adaptation planning. Its open-source nature is accelerating international research on environmental change.
SAM 2 (Segment Anything Model 2) is Meta’s next-generation model for universal image segmentation. USRA is operationalizing use of this model with a fine-tuned version for segmenting water in satellite and airborne remote sensing imagery for USGS use cases.
Panel:
Dr. David Bell, Director of Universities Space Research Association’s Research Institute for Advanced Computer Science (RIACS)
Ms. Disha Sindham , Researcher, Data Science, Universities Space Research Association at NASA Ames
Dr. Sujit Roy, Principal Research Scientist, University of Alabama in Huntsville
Related presentations:
IN43A-06 Learning Global Geostationary Representations with DINO-MAE: A Self-Supervised Foundation Model for Longwave IR Imagery
IN41C-VR8983 Context Injection via Vision-based Adapter for Finetuning Geospatial Foundation Model
IN41C-0374 Evaluation of Prithvi EO-2.0 on High Resolution Damage Assessment
U42C-02 Prithvi-EO: An Open-Access Geospatial Foundation Model Advancing Earth Science through Global Collaboration (Invited)
Contact: Dr. Suraiya Farukhi, Universities Space Research Association, Sfarukhi@usra.edu, 443-812-6945
From Pixels to Earth Action: Operationalizing Geospatial AI Foundation Models
Universities Space Research Association
Our scientists will focus on recent efforts to operationalize geospatial AI foundation models for Earth Action, including the Prithvi-EO-2 foundation model which received the 2025 AGU Open Science Recognition Prize and Meta’s SAM2 foundation model for image segmentation that USRA is using in combination with remote sensing data for USGS use cases.
Prithvi’s development, led by NASA with collaborators from the geospatial AI community including USRA, focuses on high-resolution, multisensor satellite data. It is optimized for sustainability and climate applications—such as wildfire risk assessments, flood mapping, ecosystem monitoring, and climate adaptation planning. Its open-source nature is accelerating international research on environmental change.
SAM 2 (Segment Anything Model 2) is Meta’s next-generation model for universal image segmentation. USRA is operationalizing use of this model with a fine-tuned version for segmenting water in satellite and airborne remote sensing imagery for USGS use cases.
Panel:
Dr. David Bell, Director of Universities Space Research Association’s Research Institute for Advanced Computer Science (RIACS)
Ms. Disha Sindham , Researcher, Data Science, Universities Space Research Association at NASA Ames
Dr. Sujit Roy, Principal Research Scientist, University of Alabama in Huntsville
Related presentations:
IN43A-06 Learning Global Geostationary Representations with DINO-MAE: A Self-Supervised Foundation Model for Longwave IR Imagery
IN41C-VR8983 Context Injection via Vision-based Adapter for Finetuning Geospatial Foundation Model
IN41C-0374 Evaluation of Prithvi EO-2.0 on High Resolution Damage Assessment
U42C-02 Prithvi-EO: An Open-Access Geospatial Foundation Model Advancing Earth Science through Global Collaboration (Invited)
Contact: Dr. Suraiya Farukhi, Universities Space Research Association, Sfarukhi@usra.edu, 443-812-6945
