I have been following this development for some time (see here, here, and here). In late 2009 Google introduced an App called Google Goggles. Basically, you take a picture of something, and Google tells you what it is. Take a picture of a painting, Google identifies it. Take a painting of a book, and Google shows you the Amazon.com page for the book. It’s pretty cool. But what about taking a picture of a person? Will Google tell you who that person is? Well Google may be working on such an app.
Earlier today CNN reported:
Google is working on a mobile application that would allow users to snap pictures of people’s faces in order to access their personal information, a director for the project said this week.
In order to be identified by the software, people would have to check a box agreeing to give Google permission to access their pictures and profile information, said Hartmut Neven, the Google engineering director for image-recognition development.
“We recognize that Google has to be extra careful when it comes to these [privacy] issues,” Neven told CNN in an exclusive interview. “Face recognition we will bring out once we have acceptable privacy models in place.”
Engadget has an update, which includes Google’s quasi-denial:
Google has reached out to clarify that there are no plans to introduce functionality of this sort yet, not without “a strong privacy model in place.” More importantly, however, the linking of facial recognition to personal data is described as “inventions of the reporter” rather than something the company’s actively pursuing.
I’m not persuaded that this app is purely the “invention of the reporter.” Numerous Google executives have discussed this product for years.
Google executives are wrestling over whether to launch controversial facial recognition technology after a barrage of criticism over its privacy policies.
Mr Schmidt said: “Facial recognition is a good example . . . anything we did in that area would be highly, highly planned, discussed and reviewed. When you go through these things, you review your management procedures.”
Marissa Mayer, the vice-president of Google’s search product and user experience, said: “We are blocking out people’s faces if people try to use Google Goggles to search for information about them.
“Until we understand the implications of the facial-recognition tool we have decided to block out people’s faces. We need to understand how this tool affects people’s privacy and we cannot change that decision until we do.”
The message seems to be that Google has the power to do this (they already search photos in their Picassa service), but are just looking for the right way to do this without violating privacy.
I’m not consoled by this. Particularly in light of Google’s recent showdown with the FTC over Buzz’s privacy violations, how much faith do we have that Google will in fact launch this with all privacy considerations taken care of?
This product fulfills a prediction I made way back in 2008 in Omniveillance:
With the advent of photo-sharing Internet sites like Flikr, MySpace, and Facebook, people can now upload photographs and “tag” a specific person’s identity in the photo with metadata, as if they were captioning it in a scrapbook (i.e., John Doe is the third person on the left). Although currently the tagging process must be done manually, new facial recognition such as Google’s Picasa system utilizes artificial intelligence computers to automatically index and tag the subjects of photographs.147 Software like Polar Rose is capable of scanning the entire World Wide Web, matching faces with previously tagged photos based on similarities in biometric features, and automatically tagging the photo with the person’s identity.148 Berners-Lee mentions tagging as one of the key prerequisites to the semantic web.149
Once an image is tagged, these captions can be searched and indexed like any other document on the Internet. As a result of this emerging image-analysis technology, a search engine like Google can easily correlate a person’s face with his name, contact information, personal preferences, friends, and any of his personal information located on the Internet. In fact, Google’s Director of Product Management, R.J. Pittman, “said that Google is developing visual crawling software that can be used for facial recognition and scene analysis.”150 Applied to Street View, this future technology can be combined with tagging and advanced image search capabilities to identify anyone who is recorded by omniveillance.
Omniveillance is on its way.