Microsoft recently announced that it had released 12 million building footprints in Canada to the OpenStreetMap community. The dataset was consolidated from combining StatsCan’s Open Database of Buildings with extracting footprints from aerial and satellite imagery by leveraging artificial intelligence:
In order to identify every single building in Canada, from coast to coast to coast, through machine learning, the solution was able to identify shapes, learn which were buildings and then convert everything to geometry data that is easily processible for geography applications. Canada is only the second country where Microsoft has undertaken this initiative after the U.S. and the machine had to relearn and adjust for Canada’s vast rural, agricultural and differing landscape and biomes. With adjustments along the way, the whole process took approximately four months to map every building.
This is Microsoft’s second countrywide release of building footprints as open GIS data. Last summer, Microsoft released 125 million building footprints covering the United States.
- Putting People on the Map: Microsoft launches Building Footprints in Canada, Microsoft News Center Canada