Facebook is currently working on a technology that will enable them to identify individuals in photos even if they are positioned in an inconvenient angle like looking away from the camera. The largest social media platform's new algorithm is now being tested by its Artificial Intelligence team.
During the Computer Vision and Pattern Recognition event which occurred earlier this month, the facial recognition invention was disclosed. The new technology works by matching the subject photo with other photos. By distinguishing the similarities like dressing style, body form, hairstyle and posture, the person can be identified.
The Facebook AI team submitted a study explaining that, "In the absence of a clear, high-resolution frontal face, we rely on a variety of subtle cues from other body parts, such as hair style, clothes, glasses, pose and other context."
The paper which centers on the facial recognition was introduced at the Computer Vision and Pattern Recognition conference.
Yann LeCun, leader of Facebook's AI team demonstrated the algorithm at the conference by testing it on Mark Zuckerberg. LeCun showed how the program recognizes the physical appearance, clothing and fashion of the platform's chief executive.
The team leader explained that there are many hints available for them to utilize.
"People have characteristic aspects, even if you look at them from the back. For example, you can recognise Mark Zuckerberg very easily, because he always wears a gray T-shirt," LeCun said.
The AI team gathered 40,000 photos of individuals from Flickr. The photos were of different qualities and angles. Some were totally visible while others were not. There are faces among the photos that are covered by the individuals' hair.
Under the mentioned conditions, the photos were tested on the facial recognition technology, 83% accuracy was obtained. This indicates the potential of the algorithm in identifying individuals.
