ClearSky Vision Platform

ClearSKY Vision is an AI platform that produces cloud-free, Sentinel-2-compatible imagery by fusing optical and SAR observations. Deep-learning models (Stratus-3, Stratus-2-Fast, Stratus-2, Nimbus and the on-request Cumulus) reconstruct cloud-covered, cloud-shadowed and missing pixels rather than masking them, delivering 10-band Level-2A imagery that is interchangeable with original Sentinel-2 products. Required inputs are Sentinel-1 C-SAR and Sentinel-2 MSI, with Landsat 8/9 (OLI/TIRS) required by Stratus-3 and an optional input group elsewhere, plus optional CBERS-4/4A and Amazonia-1 on Cumulus. Outputs are delivered as Cloud Optimized GeoTIFFs and XYZ tiles through a REST API and web dashboard, with a per-pixel Usability Score included, supporting GIS, analytics and monitoring workflows in agriculture, forestry, urban and environmental programmes. Only current and past observations are used, so historical series never leak future data. ClearSKY Vision is a 2023 CASSINI Challenges product winner and delivered more than 250 million km2 of imagery last year.