Even if you've never heard of "Byzantine," you can probably tell a Byzantine church from a Gothic one. Judging style differences is nearly impossible for a computer, however, and researchers from the University of Massachusetts want to fill in that gap. They used geometric matching, crowdsourcing and machine learning to teach an algorithm how to spot similar styles in buildings, furniture and other objects. That's something that could be incredibly useful for historians with mountains of photo archives, or game designers who need to auto-fill a level with historically accurate furniture.
After surveying around 2,500 people and checking over 50,000 responses, the team determined that users were 85 percent consistent in categorizing styles. With that info in hand, they developed an algorithm that compared the shapes of objects, distance components and matching elements. Spotting matching elements like domes on a church was the hardest part, so the group used crowdsourced responses from human participants to teach the machine.
The result? The program is nearly 90 percent efficient in spotting stylistic differences, on par with humans. Researchers used it to separate buildings into distinct historical styles labeled by experts as Gothic, Baroque or Asian, to name a few. The algorithm can then apply those labels to other structures, greatly automating the identification process. It could also be used for film special effects or game design to help artists quickly and accurately populate scenes. And yes, it could potentially do facial recognition so that machines can tell Sarah Connor apart from, say, Sara O'Connor.