Artificial intelligence foundation models promise to revolutionize various fields, including Earth observation (EO) and Earth sciences. Foundation models are trained on vast amounts of unlabeled data, allowing them to capture complex patterns and inherent information within the data. Once pre-trained, foundation models can be adapted to numerous downstream applications with little to no additional training samples. This paradigm shift will transform industry, research and development, and the broader scientific community.

Incorporating foundation models into the Earth observation and Earth sciences ecosystem holds great promise for enhancing the analytical and predictive capabilities of geospatial applications. However, this comes with challenges such as setting up a robust AI architecture, creating benchmarks, and adapting foundation models to the unique characteristics of EO data.

Jointly hosted by NASA and ESA (European Space Agency), the International Workshop on AI Foundation Models for EO aims to address these challenges by bringing together experts from the AI, geospatial science, climate, weather, and Earth observation communities. Topics to be covered include data curation, advancements in AI architecture, fine-tuning, model evaluation, and the challenges of integrating AI with geospatial digital twins. Additionally, the workshop will explore how foundation models can be leveraged in the context of commercial Earth observation products and services, driving innovation in the private sector.

The conference is in-person only at the ESA Center for Earth Observation (ESA-ESRIN) in Frascati, Italy, but it will be streamed. Abstract submissions are due by January 31, 2025, at 11:59 p.m. CET. Registration opens March 2025.