Particularly poor geometry around the Chicago Loop

[dhodcz2]

Although we have seen media showcasing otherwise decent geometry offered by Microsoft Building Footprints, the geometries near Chicago from the 2022 dataset seem to be particularly inaccurate: This is still not near as egregious as the chaotic geometry of this area from the 2019-2020 US Building Footprints:

[andwoi]

Thanks for highlighting this issue for Chicago. We're aware of the problem and are actively working to improve our models and processes to improve dataset generation. Stay tuned!