Facebook’s experimental facial recognition software identifies people using visual clues
Facebook head of artificial intelligence Yann LeCun said the company has been developing an experimental facial recognition technology that can use various visual clues in photos to identify individuals, according to a report by Gizmodo.
LeCun and his team recently demonstrated at the Computer Vision and Pattern Recognition in Boston how the software can look for other clues like hair color, clothing, pose, and posture to recognize a person in an image.
“There are a lot of cues we use. People have characteristic aspects, even if you look at them from the back,” LeCun said.
Using these visual clues, the Facebook algorithm is able to identify individuals, with great accuracy, most of the time.
In a recent test using 40,000 Flickr images, Facebook’s facial recognition algorithm was able to correctly identified people 86% of the time.
If Facebook’s experimental software is as advanced as LeCun claims, it could seemingly identify and locate individuals in a photo – even if they only partially appear in the background.
Previously reported, Facebook recently rolled out a new service called “Moments” that expands the use of its “faceprint” technology, however, the feature will not be launched in Europe any time soon.