FZJ-IEK3-VSA / PASSION

Software package for the estimation of rooftop photovoltaic potential with satellite imagery
MIT License
15 stars 1 forks source link

Bing Global Building Footprint #2

Open l-kotzur opened 1 year ago

l-kotzur commented 1 year ago

Hey @rguitar96

What do you think about https://blogs.bing.com/maps/2023-06/Bing-Maps-Global-Building-Footprints-released

Is it it worth to implement?

Cheers, leander

rguitar96 commented 1 year ago

Hey @l-kotzur!

It is definitely impressive work! For reference, the final data product is found here: https://planetarycomputer.microsoft.com/dataset/ms-buildings and evens comes with a nice, fast web visualization as well as a sample notebook.

I think this is very worth to implement as an alternative method in the rooftop segmentation, where currently the two options are using our own models (segmentation/prediction.py) or using OSM (segmentation/osm.py). If we build a module that generates masks in the same format as the other two, this would integrate seamlessly.

I will be quite busy during the next weeks (we need to talk!), but if I can get my hands on this I could give it a try :)