Facebook’s DeepFace Project Nears Human Accuracy In Identifying Faces

Facebook has reached a major milestone in computer vision and pattern recognition, with ‘DeepFace,’ an algorithm capable of identifying a face in a crowd with 97.25 percent accuracy, which is pretty much on par with how good the average human is (97.5 percent accurate) at recognizing the faces of other walking, talking meat sacks.

To get past the limitations of ordinary facial matching software, Facebook’s researchers have managed to find a way to build 3D models of faces from a photo, which can then be rotated to provide matching of the same face captured at different angles. In the past, facial recognition via computer could be pretty easily foiled if a subject is simply tilting their head in a slightly different direction.

The Facebook DeepFace algorithm needs to be trained on an extensive pool of faces to be able to perform its magic, but it can identify up to 4,000 identities based on a database of over 4 million separate images in its current version. Theoretically, that could be expanded to cover a much larger swatch with further work, and then be applied to Facebook’s social network itself, which would be very useful if Facebook wanted to automate the process of identifying all your contacts, and performing analytical magic like determining who you’re photographed most frequently with, without the use of manual tagging.

So far, this project is being put forward as mostly an academic pursuit,in a research paper released last week, and the research team behind it will present its findings at the Computer Vision and Pattern Recognition conference in Columbus, Ohio in June. Still, it has tremendous potential for future application, both for Facebook itself and in terms of its ramifications for the field of study as a whole.