Social media giant Facebook can currently identify its users in pictures just by examining their facial features. But what if the face is not entirely visible?
Facebook's artificial intelligence lab may have already found the answer to this question. The company has come up with an experimental algorithm so advanced that it can recognise persons through their body shapes, clothes, hairstyles and even their poses.
Researchers from this lab collated some 40,000 public photos from Flickr to test this new algorithm. Some of the pictures have faces that are fully and clearly visible, while some have the subjects facing away from the camera.
The test showed that the new algorithm that Facebook has developed can identify persons in pictures with 83 percent accuracy, even with their faces hidden from view.
Yann LeCun, head of Facebook's artificial intelligence lab, explained that his team wanted to develop a way to identify users even more advanced than the current face recognition feature of the social networking site.
"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 recognise Mark Zuckerberg very easily, because he always wears a gray T-shirt," LeCun explained.
LeCun and his team presented their new sophisticated neural network at the Computer Vision and Pattern Recognition conference in Boston, Massachusetts, earlier this month.
This new algorithm is expected to match the one recently unveiled by Google in its Photos application, which sorts out pictures based on their content.
Not everyone, however, is a fan of these technological advances. Privacy advocates, for instance, are worried that such an algorithm that can identify someone using just their physical attributes or their clothes can raise some serious implications.
"If, even when you hide your face, you can be successfully linked to your identity, that will certainly concern people," Carnegie Mellon University's Ralph Gross, a privacy advocate, said.
