GlobalBuildingAtlas GBA.Polygon (ODbL Subset): Global Building Footprint Polygons in 5-Degree GeoJSON Tiles
- Provider
- Zhu lab, Technical University of Munich (GlobalBuildingAtlas)
- License
- ODbL-1.0 (source)
- Size
- 924 files
- Stored at
- Hugging Face (https://huggingface.co/datasets/zhu-xlab/GBA.ODbLPolygon)
- Access
- Free and public. No account with the data provider is needed.
- Category
- Earth Observation & Geospatial
- Source
- https://huggingface.co/datasets/zhu-xlab/GBA.ODbLPolygon
About this dataset
The ODbL-licensed part of GBA.Polygon, the building footprint layer of GlobalBuildingAtlas from the Zhu lab (Technical University of Munich), a global dataset of building polygons, heights and LoD1 3D building models described in an ESSD preprint (essd-2025-327). This repository holds only the polygons that come from OpenStreetMap and Microsoft GlobalMLBuildingFootprints, both ODbL, so it is distributed under ODbL; polygons from other sources and the heights and LoD1 models are on mediaTUM. 922 GeoJSON files in EPSG:4326, one per 5 x 5 degree tile, named by its geographic extent, grouped into continental folders: africa, asiaeast, asiawest, europe, northamerica, oceania and southamerica. About 462 GB; read a tile with geopandas or GDAL. Supports urban mapping, exposure and population modeling and building-height and 3D city work. Licensed ODbL 1.0; attribute OpenStreetMap contributors and Microsoft and cite GlobalBuildingAtlas (Zhu et al.). The LoD1 product is a separate live row on Source Cooperative.
How to load this dataset in Python
from huggingface_hub import hf_hub_download
path = hf_hub_download(repo_id="zhu-xlab/GBA.ODbLPolygon", filename="africa/e000_n10_e005_n05.geojson", repo_type="dataset")
print(path)Data structure
Files
Example file: africa/e000_n10_e005_n05.geojson
Analyze this dataset without downloading it
On Scigantic, GlobalBuildingAtlas GBA.Polygon (ODbL Subset): Global Building Footprint Polygons in 5-Degree GeoJSON Tiles opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.
This page is also available as Markdown and as JSON.