Philly Groups Aid AI in Detecting Gentrification

Researchers at Drexel University develop a computer vision program to reliably identify and track gentrification in Philadelphia neighborhoods.

Published on Feb. 6, 2026

Got story updates? Submit your updates here. ›

Researchers at Drexel University have created a computer vision program that can identify signs of gentrification in Philadelphia neighborhoods by integrating qualitative data from community focus groups and quantitative data from images, construction permits, and Census information. The program is believed to be the first "deep mapping" machine learning model that can accurately detect gentrification in the city, with an 84% accuracy rate.

Why it matters

Gentrification is a complex issue that can dramatically impact longtime residents of urban neighborhoods. This new AI-powered tool could help community leaders, urban planners, and researchers better understand and mitigate the negative effects of gentrification, such as displacement of long-term residents. The program aims to provide more transparent and reliable data on gentrification trends to support equitable development.

The details

The researchers worked with residents in three Philadelphia neighborhoods identified as experiencing gentrification. Through focus groups, they learned about the visual cues and building characteristics that residents associate with gentrification. Using this qualitative data, along with quantitative data from images, permits, and Census information, the team trained a neural network machine learning model called ResNet-50 to identify 1,040 data points that are visual hallmarks of "new-build" gentrification. When tested on new image pairs, the program was able to correctly identify gentrification 84% of the time.

  • The research team connected with residents and conducted focus groups from 2009-2013.
  • The team labeled over 17,000 historic images of Philadelphia neighborhoods from 2009-2013 and more recent images from 2017-2024.
  • The researchers presented their work and the gentrification identification program in the journal PLOS One in 2026.

The players

Drexel University

A private research university located in Philadelphia, Pennsylvania.

Maya Mueller

A doctoral student in the College of Engineering at Drexel University who led the research.

Simi Hoque

A professor in the College of Engineering at Drexel University and a co-author of the research.

Got photos? Submit your photos here. ›

What they’re saying

“While gentrification looks different depending on where it's happening, the people who live in those areas can identify it immediately.”

— Maya Mueller, doctoral student (PLOS One)

“Residents of these areas know gentrification when they see it. In our focus groups they said these buildings 'stick out like a sore thumb.' So, it was then our job to translate the 'sore thumb' into a list of traits that we could use to train our program.”

— Simi Hoque, professor (PLOS One)

What’s next

The researchers say the program would be improved through additional use and exposure to more varied training data. They hope the tool can help urban planners, community advocates, and others better understand and address the effects of gentrification.

The takeaway

This new AI-powered tool developed by Drexel University researchers demonstrates how integrating qualitative community knowledge with quantitative data can produce a more transparent and reliable way to identify and track the complex issue of gentrification in urban neighborhoods.