Pitt Patt was founded in 2004 as a spin-off firm built upon a decade of research into object recognition by scientists at Carnegie Mellon University. Now it's a Google property. What will the Net behemoth be able to do with Pitt Patt's technology? Almost anything to do with advanced face recognition, from video to Picasa's popular images to photos uploaded and shared via Google+. This really could change web-based everything.
Pitt Patt (for Pittsburgh Pattern recognition) developed a highly successful system for recognizing people's faces. At its core are two algorithms that recognize faces--one mostly front-on with a yaw angle of 18 degrees and one that can ID people who've tilted their faces up to 36 degrees from head-on. It's also capable of tracking people and objects--meaning it's good for video feeds, too--and has a complex API to allow for sophisticated integration into different products.
What does this mean for Google? Pitt Patt's newly refreshed front page highlights that "computer vision technology is already at the core of many existing products (such as Image search, YouTube, Picasa and Goggles)" so we can take it that the Pitt Patt algorithms will be quickly adapted into Google's tech to aid with the accuracy of face and object recognition, for things like landmarks in Goggles.
That's all well and good, but what if Goggles was wired into your Android phone's camera so it automatically tagged your friends in the metadata, using a remote look-up like Goggles' system--boosted by Pitt Patt's tech? How about highly reliable face-recognition log-ins for Android phones, tablets, or Chromebooks?
Instant, reliable face recognition could also dramatically affect the services in Google+, with automatic linking of people's profiles to images and video uploaded by other users. And here we see the germs of a novel idea to quickly create a social graph that's as complex and smart as Facebook's is: By encouraging users to enable face recognition, and working out who's most often in photos and videos together, Google could almost certainly map a complicated network of friendship relationships by inferring them, and thus "steal" Facebook's biggest jewel without any actual theft.
There's a massive buzz kill (pun intended) underlying all these suggestions, however. And that comes from Google's own Eric Schmidt who has recently revealed that Google's shied away from face recognition primarily because it has too many implications for privacy abuses (and, presumably, lawsuits aimed at Google). But what if Google's taking some of the lesson's it's learned about user privacy since the Buzz and Wave fiascos--and subsequent success of Plus--and is now ready to take a limited leap into more ubiquitous automatic face recognition with privacy lessons intact? Wound throughout Google's extensive tech offerings, face recognition could become everything from a key (if your face is tied to your Google profile) to a powerful search booster. It's also tech that Apple doesn't have provision for on the iPhone or iPad--and so it could give Android a big advantage if consumers like it.
No comments:
Post a Comment