Facebook’s facial recognition abilities are advancing at a scary pace.
Example A: DeepFace, Facebook's project that can identify a face within a crowd with 97.25 percent accuracy. Facebook praises this, pointing out that this is as correct as the average human eye is at recognizing faces.
What Facebook has been able to do with DeepFace is create 3D models of faces from a user’s photo, which can be moved around to view the person at different angles. So forget tagging: the social network could one day correctly ID you from any photo that’s uploaded by anyone. DeepFace is far more powerful than the tagging system that the company uses now: it uses “deep learning,” a sector of A.I. that specializes in understanding irregular types of data.
Here’s an example of the program recognizing ol' Stallone:
As of now DeepFace is only a research project, but that doesn’t rule out it (or parts of it) being used by the social network in the future. Hey, there’s more than a billion users on Facebook, with far more photos having already been uploaded. And there will be even more tomorrow, and the day after next—just more fuel for the fire.
[via TechCrunch]
