Shock of recognition

Bob Holman and Sparrow. (photo by Jeff Wright)

Bob Holman and Sparrow. (photo by Jeff Wright)

It started when I posted a picture of my husband, Sparrow, in a Santa suit, and Facebook voluntarily inserted his name into the message. I was creeped out. Facebook had recognized my husband!

A quick Google search revealed that facial-recognition software has been used by Facebook on and off since 2011, but law enforcement has been using the technology much longer to help identify people caught on surveillance cameras. (Casinos use it too.) The software’s success rate — which requires not only sophisticated algorithms but also a database of facial images — is increasing as our handheld devices and social media boost the public stock of personal photos.

In other words, if you’re a criminal, back off on the selfies. Of course, the definition of the word “criminal” varies, so as the government’s surveillance habits are exposed by Edward Snowden, some people might get a little anxious about the ease with which we can be visually identified online. If the government is monitoring email and phone calls, what’s to stop them from mining Facebook files?


On Facebook, the idea is that automatic facial recognition saves the user the trouble of tagging friends to let them know that photos of them have been posted. The user can easily delete the tag if desired, but the people so tagged have no control over their IDs flying out across the site. You can opt out of being tagged by clicking “Account Settings,” then “Timeline and Tagging,” for a list of options. Before you follow through, you’ll get a message imploring you to leave tagging on as an aid to “sharing memories and experiences with friends.”

The good news, at least at this point, is that the recognition software is still pretty crude. I decided to experiment by posting other pictures of my husband and myself. None of them elicited a response from Facebook, except when I tried again with the Santa pic. By scrolling back through my husband’s timeline, I found a photo a friend had posted earlier in which both Sparrow and poet Bob Holman had been tagged, apparently by the person posting. Sparrow’s head was tilted in almost the same position as in the Santa picture, and he had a similar expression on his face. Perhaps this post was the source of Facebook’s identification.

Full-face frontal images are easiest for the software to match up, but of course, the photos we take on Smartphones are often blurry or taken at odd angles. Animetrics, Inc., a facial-recognition software developer, has invented algorithms that can convert a photograph into a set of 3D images, rotating the head to improve the chances of making a match among a database of pictures, which may be posed at angles other than full-face. Some types of Animetrics software are already running on the Smartphones of police officers.

In 2011, Samsung first offered its Android phones with a Face Unlock feature, using facial-recognition technology to allow only the phone’s owner to unlock the phone. When it was shown that a photograph of the owner worked just as well, Samsung added blink-detection programming. While unlocking the phone, the user has to blink to prove she’s not a photo. According to the TechNewsDaily website, Google has filed a patent on a technology that would require the user to assume a goofy expression to unlock her phone.

Google’s embrace of facial recognition took a step backward when members of Congress expressed their concerns about including the technology in Google Glass, a product currently under development that the company hopes will revolutionize modern life. Google Glass is a wearable computer that perches on the nose and ears like glasses. It contains a microphone, a camera, an Internet hookup, and more. It responds to voice commands so the user can take photos or videos of whatever he’s looking at, hands-free. (But can it do selfies?)

At this point, Glass is available only to developers, who pay $1500 for the privilege of experimenting with the device and writing programs that will help apply its use to such varied needs as translating signs from foreign languages, recording golf scores and course data (distance from tee to hole, for instance), obtaining on-the-ground directions from GPS, and sharing the current panorama via social media. When Congress expressed concerns that Glass would produce privacy violations of all kinds, Google’s response was to announce that it would not include facial recognition in its product — at least not yet.

With the prospect of Glass on the horizon, and a future in which people can make videos of us just by staring and saying, “Take a video,” I suppose a little Facebook ID popping up doesn’t seem like such a big deal. And like most technology that proves useful, entertaining, or time-saving, it probably won’t be long before facial recognition is an accepted tool of modern life — for better or worse.