Facebook Creates Software That Matches Faces Almost as Well as You Do


Facebook’s new AI research group reports a major improvement in face-processing software.

Asked whether two unfamiliar photos of faces show the same person, a human being will get it right 97.53 percent of the time. New software developed by researchers at Facebook can score 97.25 percent on the same challenge, regardless of variations in lighting or whether the person in the picture is directly facing the camera.

That’s a significant advance over previous face-matching software, and it demonstrates the power of a new approach to artificial intelligence known as deep learning, which Facebook and its competitors have bet heavily on in the past year (see “Deep Learning”). This area of AI involves software that uses networks of simulated neurons to learn to recognize patterns in large amounts of data.

“You normally don’t see that sort of improvement,” says Yaniv Taigman, a member of Facebook’s AI team, a research group created last year to explore how deep learning might help the company (see “Facebook Launches Advanced AI Effort”). “We closely approach human performance,” says Taigman of the new software. He notes that the error rate has been reduced by more than a quarter relative to earlier software that can take on the same task.

Facebook’s new software, known as DeepFace, performs what researchers call facial verification (it recognizes that two images show the same face), not facial recognition (putting a name to a face). But some of the underlying techniques could be applied to that problem, says Taigman, and might therefore improve Facebook’s accuracy at suggesting whom users should tag in a newly uploaded photo.

However, DeepFace remains purely a research project for now. Facebookreleased a research paper on the project last week, and the researchers will present the work at the IEEE Conference on Computer Vision and Pattern Recognition in June. “We are publishing our results to get feedback from the research community,” says Taigman, who developed DeepFace along with Facebook colleagues Ming Yang and Marc’Aurelio Ranzato and Tel Aviv University professor Lior Wolf.

Written By: Tom Simonite
continue to source article at technologyreview.com


  1. “It was a bright cold day in April, and the clocks were striking thirteen.

    The beginning was prosaic, disarming, and simple. Little did we know that, by the end, it would predict the moment when humans, once and for all time, decided to trade their Freedom for Happiness.

    Well, enough about Facebook. As for 1984,…

  2. In reply to #3 by PERSON:

    I guess this has the potential to make the GCHQ webcam database accessible to staff of all sensibilities (and perhaps more positively, help restrict access to at least genuinely similar looking people).

    Thanks for the link to GCHQ, I hadn’t heard about that. The story was interesting because for once it’s the US citizens who may be getting impacted by the “our security forces can fuck with people’s privacy as long as those people aren’t our citizens” mentality that is so prevalent in the US.

  3. Could it detect plastic surgery in those creepy avatars as well as humans could ?
    Facebook – share all your private thoughts publicly…

  4. I saw a news show about this some time ago. Leslie Stahl walked into a restaurant and a camera near the door “scanned” her and sent a coupon to her phone within about a minute. This idea is troubling to me, so much so, that I don’t ever want my face on the internet.

  5. I like the coupon angle, but it scares me to think about child predators using this technology to snap a child’s picture, scan the internet and gain their trust by calling them by name.

Leave a Reply