Using a revolutionary new AI-powered method, researchers are mining old insurance maps to visually reconstruct "ghost" neighborhoods — with promising results. The process could help with efforts to better estimate the economic loss caused by the demolition of historically diverse neighborhoods nationwide. It could also humanize that oft-forgotten history, bringing it to life in a way that raw data and spreadsheets simply cannot. The technique involves using machine learning algorithms to mine digitized 19th- and 20th-century fire insurance maps — effectively creating 3D computer models of long-lost places. Using the technique, researchers can pinpoint a since-destroyed building's location and footprint, plus figure out how many floors it had, the construction materials used, and its overall purpose. The system harnesses Sanborn Fire Insurance Maps, which are highly accurate and were originally created to gauge American cities' fire liability. The method can be used to develop 3D digital models for any of the roughly 12,000 American cities and towns with Sanborn maps. "Now with machine learning, it is really possible to analyze all of that information at scale ... and that has not been really possible in the past," says study lead author Yue Lin, a former doctoral student in geography at Ohio State who specializes in spatial data science. Miller's team at the Center for Urban and Regional Analysis is working on a 3D digital model of one of the country's first public housing projects, as well as digitally reconstructing a Columbus avenue that housed Black-owned businesses before highway infrastructure reshaped the community. Miller expects the 3D visualizations to be accessible as virtual reality photorealistic immersion experiences within a year — people will be able to put on a headset and "walk down the street in one of these neighborhoods," he says. The team is also in talks with the Philadelphia Federal Reserve Bank about applying the method to help measure historic economic activity generated by the city's communities of color that were disrupted by urban renewal.
