Facebook's Deep Face facial recognition technology to have near-human accuracy

PALO ALTO, Calif., March 19 (UPI) -- Facebook is developing facial recognition technology that will recognize faces with 97.25 percent accuracy as compared to 92.75 percent for humans.

Deep Face, which is being developed by the social network giants' API department, promises near-human accuracy and could mean users won't have to tag their friends in pictures. The technology will create a 3D map of facial features and create a colorless model to narrow in on specific characteristics.

"We present a system (DeepFace) that has closed the majority of the remaining gap in the most popular benchmark in unconstrained face recognition, and is now at the brink of human level accuracy," researchers said in a report released by Facebook API Group.

The API team used 4.4 million tagged faces from 4,030 Facebook users to help the system learn and improve its recognition capabilities.

According to Facebook, facial recognition consists of four stages: detect, align, represent and classify. Facebook is using their 3D modeling technology to revisit both the alignment and the representation steps. They then recognize the face using a nine-layer deep neural network.

There is no confirmation as to when Facebook will release the technology, but the company will present it at the IEEE Conference on Computer Vision and Pattern Recognition in June.

[Facebook] [Mashable]