Multiclass Road Surface Dataset from Satellite Imagery

20260 citationsDatasetgreen Open Access

Authors

FORREST MUMBERE MUYISA · Sharda University

VERMA Dr. Prof. SHIV KUMAR · Sharda University

Abstract

# 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

Topics & Keywords

Publication Details

Published in: Zenodo (CERN European Organization for Nuclear Research)

DOI: 10.5281/zenodo.19364351

Command Palette

Multiclass Road Surface Dataset from Satellite Imagery

Authors

Abstract

Topics & Keywords

Publication Details