Facebook’s new Algorithm powerful enough to identify you even if you aren’t looking at the camera.
Facebook’s AI/artificial intelligence lab introduced an algorithm at the Computer Vision and Pattern Recognition conference earlier this month that can locate you in a picture with 83% accuracy.
The catch is that it can judge your identity even if you aren’t looking at the camera or have turned your face away.
PIPER (Pose Invariant Person Recognition) is the name given to the identification method. PIPER uses poses, identifying features, body shape, clothing, hairdo, partially exposed faces and similar subtle clues, which otherwise can’t be computed by facial recognition technology.
Facebook’s AI team’s head Yann LeCun explains: “There are a lot of cues we use. People have characteristic aspects, even if you look at them from the back. For example, you can recognize Mark Zuckerberg very easily, because he always wears a gray T-shirt.”
The team took around 40,000 random public pictures from Flickr including the photos where faces were turned away and passed them through an advanced neural network. Resultantly, the algorithm was able to identify identities of different people with 83% accuracy. This sort of an algorithm can help greatly in enhancing the efficiency of photo apps like Facebook Moments.
The accuracy of this Algorithm is 83%
The technology developed by Facebook is definitely impressive as well as a powerful but naturally some users would think of it as creepy. Included in such users are other AI field academics.
The Robotics Institute at Carnegie Mellon University in Pittsburgh, Pennsylvania’s postdoctoral fellow Ralph gross told New Scientist, “If, even when you hide your face, you can be successfully linked to your identity that will certainly concern people. Now is a time when it’s important to discuss these questions.”
Apparently, Facebook is doing whatever it can to perfect this new technology and regardless of the criticism they will be improving it.