San Diego State University is using artificial intelligence to track the migration patterns of unsheltered homeless people in the county.
San Diego County measures its homeless population with an annual Point-in-Time Count.
“But we know that the homeless population is actually very dynamic,” said project lead Ming Tsou. “They’re moving every single day.”
Tsou is a professor in the Department of Geography at San Diego State University Global Campus.
His project combines city, county and nonprofit data with a new approach – using AI to identify tents in aerial and streetview images, similar to what can be seen on Google Maps. The tool estimates how many people live inside those tents based on their size.
He said the result is a more comprehensive and real-time measure of where San Diego’s unsheltered homeless population lives and where and how they move. The information could help local governments and nonprofits decide where to place resources like shelters, handwashing stations, and street medicine teams.
Tsou’s team uses something called geomasking to protect the people they’re tracking – displaying population densities or hotspots without revealing exact locations as pinpoints.
The project relies on artificial intelligence, but it’s also very human.
His students drive around to see where AI is making errors and why.
They survey the people living in tents, asking questions like – why stay here? Where did you stay yesterday and why did you leave?
Tsou is adamant that they must lead with empathy.
“When we analyze the data it’s just a number. Digital numbers, digital maps. But every single number, behind that is a story of an unhoused individual,” he said.
He plans to one day use the model to make predictions and help communities prepare before homeless people move.
He hopes policymakers will use the information to support not just homeless people, but the neighborhoods they’re gathering in, which his team’s data show tend to be lower-income, less educated and “more vulnerable,” he said.
