Stereoscopic Image Analysis is a web application that uses computer vision and machine learning to help identify the unmarked graves of Civil War soldiers in St. Augustine, Florida. Sponsored by the National Cemetery Administration, the project addresses a genuine historical mystery: twenty men of the 7th New Hampshire Volunteers died in St. Augustine during the war, and while old stereoscopic photographs clearly show nine of their names on wooden headboards, those markers rotted away before being recorded—and in 1905 the graves were mistakenly relabeled as unknown soldiers. The application analyzes pairs of these 19th-century stereoscopic photographs using epipolar geometry and machine learning to calculate real-world distances within the images, even when the original camera’s specifications are unknown. By measuring those distances and matching them against the present-day site, the tool helps pinpoint exactly where each soldier is buried—offering a non-invasive way to restore the names of the fallen, bring closure to their descendants, and preserve a piece of American history.
