The Cornell AI & Vision for Earth group focuses on building computer vision and machine learning techniques for earth observation, with an eye towards earth science and environmental science applications. We build foundation models as well as interpretable techniques for using these models to answer any questions about the earth.
Our multimodal dataset for disaster monitoring has been accepted as a Spotlight paper at the NeurIPS Datasets and Benchmarks Track.
Our benchmark for evaluating LLM agents on Earth observation tasks was presented at the ICML workshop.