CrisisLandMark

CrisisLandMark is a large-scale, multimodal corpus for Text-to-Remote-Sensing-Image Retrieval (T2RSIR). It contains over 647,000 Sentinel-1 (SAR) and Sentinel-2 (multispectral optical) images enriched with structured textual and geospatial annotations. The dataset is designed to move beyond standard RGB imagery, enabling the development of retrieval systems that can leverage the rich physical information from different satellite sensors for applications in Land Use/Land Cover (LULC) mapping and crisis management.

Source: Hugging Face Hub (DarthReca/crisislandmark). Metadata imported from the dataset’s Hub tags.