< BackHome/news/facial-recognition-civil-war-photos

History Lessons

It’s estimated that there are more than 40 million photographs and portraits from the American Civil War, but only 10 to 20 percent have people that are identified in them. The Civil War Photo Sleuth is aiming to put an end to that by applying facial recognition to photographs from that era so that similar pictures can be grouped together. If some of those images contain a subject’s name or signature, then that can be easily added to the other non-tagged images. Crowdsourcing also comes into play: Anyone is invited to help identify people in the images on the project’s website, as well as contribute any of their own photos. The point is finding them all, which is not as easy as you’d think: Today’s facial recognition tools can be challenged by the beard and sideburn fashions of the day of the day and the side profile view portraits that were popular back then, but this will certainly improve with more time and training. AI-optimized archival photos are getting a lot of play lately. Google and the New York Times recently unveiled a project to digitize the newspaper’s entire photo archive, which will make it easier to find many photos that were never published, for example, thanks to the tech giant’s object and optical character recognition tools.
View Full Article >

sign up for our newsletter