When it comes to recognizing the subject of a photo, human beings are 97.53 percent accurate. That’s pretty good, but the facial recognition software that Facebook developed, DeepFace, is almost as accurate. That system accurately recognizes people 97.25 percent of the time.
How does it work? First, DeepFace identifies your features among the nearly 400 million photos uploaded to Facebook every day. Users will then get an alert from Facebook notifying them that they have been tagged in a picture. From there, they can choose to blur their face out of the photo. (The system is already in place, but it only reveals the identities of your friends.)
As sci-fi as this technology sounds, it’s very real — and could pose very real privacy problems. And its potential users are virtually limitless; advertisers could identify the faces of in-store shoppers to better tailor their deals, and people could be similarly analyzed via smart phone cameras or wearable tech like Google Glass. And that’s not even mentioning the fact that DeepFace automatically tags users on Facebook without them having a say in the matter. The only privacy setting addressing this issue is located in the Timeline and Tagging section. You can set the ‘tag suggestions’ to no one.
There’s no doubt that facial recognition technology is a nearly unprecedented innovation, and one of the most difficult tasks yet that a computer has successfully completed. However, it ultimately treats the faces of users like just another piece of data to collect, and that could prove to be a huge privacy problem.