Multiclass Road Surface Dataset from Satellite Imagery
收藏资源简介：
# Multiclass Road Surface Dataset from Satellite Imagery ## DescriptionThis dataset provides pixel-wise annotations for road surface classification using high-resolution satellite imagery. It contains 600 images with their corresponding mask and tiff files. The shapefiles for highway, paved and unpaved have been added. ## Classes0: Background 1: Highway 2: Paved road 3: Unpaved road ## Data Sources- Satellite imagery: Google satellite- Road labels: OpenStreetMap (OSM) ## Methodology- OSM road vectors were extracted and classified based on tags: - motorway → highway - primary/secondary → paved - track/unclassified → unpaved- Data was reprojected to EPSG:32644- Rasterization performed at 512×512 tiles- Alignment ensured using GeoTIFF affine transforms ## Applications- Road segmentation- Surface classification- Disaster impact analysis- Infrastructure mapping
