Facebook's facial recognition is only 0.28% less accurate that a human's

Imagine a computer that can recognize your friend virtually as well as you can. The day is already here. Facebook tested new facial recognition software that can place faces with a 97.25% accuracy, only 0.28% less accurate than a real person’s abilities. “It’s not being used right now — I don’t want people to get up in arms yet,” said Lydia Chan of Facebook’s technology communications department.

Facebook has revealed in a research paper that it has discovered a way to match faces in two different photographs with 97.25 percent accuracy. The research has aided in the development of a software project called DeepFace which only slightly trails the facial recognition measured average of a human by only 0.28 percent. The increased accuracy comes at a rate of close to 25 percent over other software by a process of facial verification rather than recognition. Improvement in the rate is also regardless of changes in angle of the face towards a camera or differences in lighting. “You normally don’t see that sort of improvement,” says Yaniv Taigman, one of the researchers on Facebook’s AI team. 

Read full article

Comments