© 2024 WOSU Public Media
Play Live Radio
Next Up:
0:00 0:00
Available On Air Stations

How old maps and machine learning are helping to recreate lost Columbus neighborhoods

A computer representation of houses in a 1960 Columbus neighborhood that no longer exists.
Harvey Miller, Ohio State University
A three-dimensional representation of the Columbus neighborhood of Driving Park as it would have looked in the early 1960s.

A team at The Ohio State University has been turning old fire insurance maps into 3-D renderings. Geography professor Harvey Miller says Sanborn maps from the 19th and 20th centuries include a lot of information that can be fed into a computer.

"The information about the buildings would be the footprint, how many stories, what it was constructed out of, and its major use — was it a dwelling, was it commercial, or was it industrial?" he says. "It's a wealth of detailed building-level data."

Using machine learning, Miller and his team fed that data into computers, which were able to construct 3-D images of the buildings and their placement on city blocks. "In order to extract this information easily in its scale so we can do this for entire neighborhoods and even entire cities," he says.

Miller says the results can be used to build a virtual recreation of a neighborhood, or to print a 3-D model. They want to refine the process to get images that are photo-realistic.

 A computer rendering of a 20th century housing project.
Harvey Miller, Ohio State University
Poindexter Village, as rendered by machine learning in the early stages of the process.

He says the results can be invaluable to urban science and historical research. "But also, we want to create these visualizations to make the past come alive for people. For them to really, really understand what happened to our cities because of some of our policies we did in the 20th century — building highways through Black neighborhoods; redlining around some neighborhoods; and not lending money for mortgages and for property ownership."

Miller says so far, they've recreated a couple of Columbus neighborhoods.

"One is Hanford Village and Driving Park," he says. "It was mostly populated by Black people. I-70 came through in 1967 and bisected the neighborhood. And what we're doing is looking at the impact of these urban highways on these communities, how they divided and damaged them."

The team has also worked on Poindexter Village in Columbus.

Miller says the 3-D visualizations can help bring history alive in a way maps, charts and figures don't.

"When you can create a real-looking 3-D visualization, even on a web page it's really impressive, but imaging creating a VR experience where you can walk down one of these streets. That changes things. That creates a real visceral feeling. People really understand that what's there now was not what was there in past, and also what can be there in the future can be different as well."

Miller says some cities are in the planning stage or process of removing urban highways. "Rochester [New York] took out an urban highway. Syracuse is going to. Detroit just got funding to do that. I think New Orleans is considering this," Miller says. "So if we're going to take out some of these highways we have to understand what was there before so when we stitch these cities together, they're true to their past, which they were thriving neighborhoods back then. Why would we do something different?"

He says the code and processes are available to anyone who wants to tackle a project. "I think in about a year or two we'll be able to take this show on the road. We've already had a lot of interest. We've been contacted by people in Phoenix, Los Angeles, Philadelphia… some other towns around the country, all interested in this technique."

Miller says they've already done a test run with Philadelphia data on a couple of blocks, and it "worked fine."

Bill Rinehart started his radio career as a disc jockey in 1990. In 1994, he made the jump into journalism and has been reporting and delivering news on the radio ever since.