The guys from Facebook are working on improving the company's facial recognition algorithms, and it appears they've been successful. The A.I. research team managed to come up with a new technique which no longer needs to actually see your face in order to make a positive ID. This means that, in the near future, the social network will be able to correctly tag you in images even if your facial features are partially or fully covered. The company hopes to be able to embed this technology into its recently launched Moments feature as soon as possible.
Apparently, this new algorithm won't just use facial features, but also other criteria like hairstyle, body language, clothing, body shape, etc. to correctly identify the people in photos. Yann LeCun, head of artificial intelligence at Facebook, stated: “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 tested this new technique on 40,000 publicly available photos from Flickr and managed to obtain an accuracy rating of 83%, which is quite impressive.
As far as the users are concerned, this new and improved algorithm doesn't only mean better photo tagging capabilities, but also a great way of knowing when a new picture of them has been uploaded on the Internet. Many have expressed privacy-related concerns but, from my perspective, if you're on Facebook and have the auto-tagging feature enabled for your profile, you've already said goodbye to privacy.
Source: New Scientist