AI Model Spots Early Signs of Gentrification from Street View Images
-
Researchers trained an AI model to recognize visual signs of gentrification from Google Street View images. The model had 74% accuracy compared to previous methods.
-
The model found early signals of gentrification missed by past demographic data from census surveys. It allows more granular neighborhood analysis.
-
Lack of quality data has made studying displacement from gentrification difficult in the past. The model offers new potential.
-
With better gentrification data, cities could target anti-displacement policies more precisely to help vulnerable residents.
-
Researchers plan to apply the computer vision technique to study impacts of extreme weather events and tie changes in the built environment to community health.