The blogosphere was set ablaze this week with the report that Google was building a facial recognition application that would let users take pictures of people and identify them within seconds.
The problem with the CNN report is that it isn’t exactly true, or at least not yet. The real story is more nuanced than that, sources familiar with Google’s thinking claim.
To rewind, CNN reported March 31 that Google is working on a mobile application that would allow users to identify users by snapping pictures of their faces with their smartphone’s camera.
Hartmut Neven, the Google engineering director for image recognition development, told CNN people would have to check a box agreeing to give Google permission to access their pictures and profile information. This opt-in action would be one way Google solves the inherent privacy question mark that comes with such a program.
Tech bloggers read the report and ran with it, concluding that Google would draw the ire of privacy advocates and regulators already heated from Google’s WiFi spying incident and Google Buzz snafu.
Google claimed it is not introducing a mobile app for facial recognition, and argued that what the CNN article positioned as real product and process was based on hypothetical scenarios of how such an instantiation could work.
Some bloggers suggested that CNN reporter Mark Milian blew coverage on the story. CNN stuck to its story, arguing that it has an audio recording of the interview. Search Engine Land, one of the blogs that covered CNN’s initial story, published this statement from CNN:
“Google’s claims do not fit the facts of the situation. This interview was prearranged – on the record – and staffed by a Google PR rep, who raised no objections at the time and did not deny what the engineer said. Additionally, we have an audio recording of the interview, as does Google. We stand firmly behind Mark’s reporting.”
Google doesn’t dispute this characterization, but sources familiar with the situation told eWEEK what CNN failed to make clear is that there isn’t a new, specific facial recognition application in the works. Rather, Google is looking how to make facial recognition a feature that would be incorporated in Google Goggles.
CNN’s article, Google believes, failed to make this clear from the outset. What it sounds like is a classic failure to set the boundaries in a meeting between reporter and subject. Onward and upward then.
Developed by Neven and now overseen by Google Product Manager Shailesh Nalawadi, Goggles is the company’s visual search application. Goggles users can currently take pictures of book covers, wine bottles, landmarks, paintings and a few other product categories and see search results about those objects retrieved within seconds.
Goggles is the context in which facial recognition would appear, but Google told eWEEK April 1: “As we’ve said for more than a year, we will not add facial recognition to Goggles unless we have strong privacy protections in place. We’re still working on them. We have nothing to announce at this time.”
Google Goggles as Social Networking App?
Facial recognition is one of the most frequently requested Goggles features. But, as Nalawadi told eWEEK in an interview last month, Goggles requires several other improvements.
For example, Goggles needs to extend the ability to recognize 3D objects beyond landmarks, and fill out the long tail of searches in any specific product category.
Goggles also needs to get faster; some searches take several seconds as the request travels across 3G networks to Google’s cloud and back across the networks to users’ phones. Local caching in the smartphone app, similar to what Google has done with its Google Maps 5.0 for Android app, could help with this.
What no one is disputing through all of the he-said, she-said, is that Google has the technology to do this from acquiring Neven’s company Neven Vision in 2005, and then grabbing Like.com last year.
Both companies featured facial recognition capabilities, but Google has resisted implementing the technology for fear of privacy concerns. The position proved wise; 2010 would see Google weather two colossal user privacy imbroglios.
In February 2010, Google launched its Buzz social conversation service, and exposed millions of users’ Gmail contacts once they opted into the service. Users sued. Google just settled with the Federal Trade Commission.
In a far bigger deal, Google Street View collected 600GB of email, browser info and user password info from unsecured wireless networks across 30-plus countries. Google is still dealing with the fallout from this issue.
Facial recognition would certainly help Google’s social networking prospects, ABI Research analyst Mark Beccue has told eWEEK. Imagine snapping a pic of your friend’s face and then seeing their status updates from Facebook and Twitter. With facial recognition, Goggles could enable this, possibly as some sort of augmented reality overlay.
For all its caution, Google may want to step lively to this. Rumor has it Apple will be incorporating this sort of capability into its future iPhone, courtesy of its acquisition of Polar Rose last year.
Going forward, facial recognition will be a key feature in the mobile war Google and Apple are waging with one another. But facial recognition is not a separate app at Google and its launch is hardly imminent.