Open Cities AI Challenge: Drone Imagery and Building Footprints for 10 African Cities

Provider
GFDRR Labs (World Bank Global Facility for Disaster Reduction and Recovery)
License
ODbL-1.0 (source)
Format
JSON (in sampled files)
Stored at
Azure Blob Storage, https://radiantearth.blob.core.windows.net/mlhub/open-cities-ai-challenge
Access
Free and public. No account with the data provider is needed.
DOI
10.34911/rdnt.f94cxb
Category
Earth Observation & Geospatial
Source
https://doi.org/10.34911/rdnt.f94cxb

About this dataset

Very high resolution drone and aerial orthomosaics over 10 cities and regions in Africa with OpenStreetMap-derived building footprint labels, released by the World Bank's Global Facility for Disaster Reduction and Recovery (GFDRR) for the DrivenData 'Segmenting Buildings for Disaster Resilience' challenge. The imagery was collected by local mapping communities (for example in Accra, Dar es Salaam, Kampala, Monrovia, Niamey, Pointe-Noire and Zanzibar) at centimetre-scale ground sampling distances. Training data are split into tier 1 (more complete, verified labels) and tier 2 (noisier labels), and the test set holds 11,000 unlabelled 1024 x 1024 chips. Measured 2026-11-15: 23,322 files, 87.5 GB. train_tier_1/ and train_tier_2/ hold Cloud Optimized GeoTIFF scenes with GeoJSON building polygons organised by city code (tier 1: acc, dar, kam, mon, nia, ptn, znz) and STAC catalogs; test/ holds the chips as GeoTIFF with STAC JSON. Read with rasterio and geopandas; Azavea's benchmark and the winning models are on GitHub. Released under the Open Database License (ODbL 1.0); cite GFDRR Labs (2020), Open Cities AI Challenge Dataset, Radiant MLHub, doi:10.34911/rdnt.f94cxb.

How to access this dataset in Python

import adlfs
fs = adlfs.AzureBlobFileSystem(account_name='radiantearth', anon=True)
fs.ls('mlhub/open-cities-ai-challenge/')[:10]

Lists the files with anonymous access. Runs on any machine, with no cloud account.

Data structure

File types in sampled files: JSON

Files

Example file: .source/metadata.json

Analyze this dataset without downloading it

On Scigantic, Open Cities AI Challenge: Drone Imagery and Building Footprints for 10 African Cities 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.

Related Earth Observation & Geospatial datasets