A Blog by Jonathan Low

 

Aug 7, 2011

Face Time: New Recognition Techs Do It All From Surveying Bar Scenes To Establishing Bond Between TV and You

The market has spoken - again. No, not the capital markets, the market for mobile tech convenience. And what it is telling us is that privacy continues to lose when it bumps up against time saving ease of use. Venture capital funding and entrepreneurial effort continue to pour into facial recognition. But for entertainment, not security.

Given the outcry every time Facebook gets caught sneaking some new app into its service one would think that public opposition to unwanted intrusion is rife. But the key word is 'unwanted.' Facebook is an established community backloading features that its friends are not sure they want. And as the Facebook demographic ages - with cute pics of pets and grandchildren threatening to overtake blurry snaps of sorority beer chugging contests - the opposition there may grow. But for a younger, more urban profile, where voluntary downloads rule, entertainment connection is king.

The reputational and business implications remain. How does one monitor the system for abuses or, more to the point, what is the mechanism for redress once the abuses occur? For the moment, however, demand is outstripping complaint. If tattoos can be erased with lasers and online reputation consultants retained, the guess here is that the facial recognition market will continue to be robust. JL

Emily Steel reports in the Wall Street Journal:
Long the realm of science fiction, advanced technologies that identify faces now are emerging as the hottest entertainment gimmick, despite the potential for privacy concerns.

One of the latest is SceneTap, a free application for iPhone and Android smartphones released in recent weeks that displays real-time stats on the local bar scene. Based on information collected via face-detection cameras installed at participating bars, the app shows the number of people at the bar, the male-to-female ratio and the average age of patrons.

One of the latest is SceneTap, a free application for iPhone and Android smartphones released in recent weeks that displays real-time stats on the local bar scene. Based on information collected via face-detection cameras installed at participating bars, the app shows the number of people at the bar, the male-to-female ratio and the average age of patrons.

Another application called SocialCamera allows users to snap a picture with their Android phone, instantly recognize their Facebook friends in the frame, tag the photos and post them to the Web. In other games, the technology detects which celebrity a person most resembles or whether they have the genetic traits of a vampire.

And hitting the market in time for the holidays this year are television set-top boxes that include facial-recognition cameras. The technology, developed by Palo Alto, Calif.-based Viewdle will be able to identify who is sitting in front of the TV then customize programming according to that individual, displaying most recently watched or recorded shows, for instance. Parents also could program the device to limit which channels their children can access.

"We're at a tipping point where some of these face-recognition technologies are not just gimmicks but are becoming useful," says Jason Mitura, chief product officer at Viewdle. "The TV begins to entertain you because it knows who you are."

Facial-recognition tools quickly are finding their way into the hands of the general public after decades in research labs. Computer scientists had to figure out how to train computers to detect human faces within the context of photographs and videos, and then match those faces against a database to identify individuals. It's tricky work, experts say, noting that algorithms need to account for people who don't look directly at the camera as well as other variations, such as lighting and expressions.

"Our brains are just so good at understanding the visual world. Such a big part of our brains is devoted to that. It is something that is extremely hard for computers to do," says Henry Schneiderman, president and chief executive at Pittsburgh Pattern Recognition Inc., or PittPatt, which Google Inc. acquired in July.

Facial recognition has come a long way after years of false starts. On still frontal face images, the error rate of rejecting a legitimate claim—when the face image and name match—decreased to 0.29% in 2010, from a rate of 79% in 1993, according to a study by the National Institute of Standards and Technology.

Initial deployments of the technology were largely for law enforcement, security and surveillance. But as cameras grow more sophisticated at capturing and recognizing faces, and as prices for the technologies drop, sophisticated face-identification tools increasingly are accessible to the general public.

Animetrics Inc., a facial-recognition company based in Conway, N.H., that focuses on the law-enforcement and security industries, launched a free app for the iPhone called FaceR Celebrity that allows users to match their face to a star. The application, which has been downloaded about 30,000 times, uses the same facial-recognition technology deployed by local law enforcement to identify criminal suspects, says Animetrics CEO Paul Schuepp.

Indeed, the fast-growing pervasiveness of face-identification technologies raises privacy concerns. Privacy advocates question what information these companies collect about people and how it could be used. The new technologies raise fears that face information about people could be used for surveillance, and wonder whether consumers would be able to opt out of the tracking. Just this week, researchers at Carnegie Mellon University in Pittsburgh unveiled a study that revealed they could identify people in photos about one-third of the time using face-recognition technologies.

The application makers say that they don't collect or store face information about their users.

With the SceneTap app, for instance, a person walks into one of the more than 250 participating bars across the country that have installed face-detection cameras. The camera scans the person's face and instantly analyzes his or her facial characteristics. The technology then matches that person's facial structure against an anonymous database that determines a person's likely age and gender. Consumers then can check out the smart phone app to survey the local bar scene. Bars receive analytics about their customers.

The company plans to make money through selling special offers to the participating bars or restaurants that will appear on the app. SceneTap launched in Chicago and will be expanding to other cities in coming weeks.

While SceneTap doesn't ask for patrons' permission to capture their image, nobody sees the video feed, and the information isn't recorded, says Cole Harper, chief executive at SceneTap. He says that the company doesn't collect any personal information. The technology also doesn't link up to Facebook or match against publicly available photos on the Web to identify exactly who is entering the bar, or scan for more controversial characteristics like ethnicity.

"From a technology standpoint, I would be lying to say that is not possible. But there are a lot of bridges that need to be crossed," Mr. Harper says.

0 comments:

Post a Comment