dClimate Labs
TESSERA introduces a framework that generates precomputed, multimodal pixel embeddings at 10m global resolution for Earth observation data, integrating Sentinel-1 and Sentinel-2 time series. The system delivers state-of-the-art or comparable performance across five diverse downstream tasks, including crop classification and canopy height estimation, often with significantly less labeled data.
There are no more papers matching your filters at the moment.