Remote sensing dataset of disaster-affected buildings and farmland in China and Pakistan

This dataset comprises two major components: a damaged building dataset and an affected farmland dataset. (1) The damaged building dataset includes post-disaster imagery and interpretation annotation data across three disaster scenarios: earthquakes, floods, and wildfires. (2) The affected farmland dataset includes post-disaster imagery, interpretation annotation data, and corresponding shapefile data specifically for flood scenarios.The specific disaster scenarios and their corresponding post-disaster imagery are detailed as follows:1. Earthquake Scenarios2025 Xigaze Earthquake: Beijing-3 post-disaster imagery (0.75m resolution).2010 Yushu Earthquake: QuickBird post-disaster imagery (0.61m resolution).2. Flood Scenarios2025 Beijing Flood: Jilin-1 post-disaster imagery (0.5m resolution).July/September 2025 Pakistan Floods: GF-1 (Gaofen-1) post-disaster imagery (2m and 1.5m resolutions).3. Wildfire Scenarios2020 Xichang Wildfire: GF-1 (Gaofen-1) post-disaster imagery (2m resolution).The acquired post-disaster imagery underwent preprocessing, including image rectification, coordinate transformation, and georeferencing. Subsequently, the imagery was combined with vector data regarding damaged buildings and affected farmland, which were obtained through manual visual interpretation. The final dataset was constructed following cropping, screening, and classification. Within the dataset, both the post-disaster imagery and interpretation annotation data are provided in .png format, standardized to a size of $256 \times 256$ pixels. The Shapefile data is compatible with mainstream GIS software such as QGIS, ArcGIS, and GDAL. The current dataset is complete with no missing entries; any future data additions will be noted in the update specifications.