Read original story on: The Next Web
Facebook’s facial recognition algorithm was first developed to enable easier “tagging” in photos to facilitate party photo sharing on the social network. Fast forward to now, however, the tool has become so powerful that it can recognize users in pictures when their face is not in shot. Instead, it identifies people based on their hair, clothing style, and other physical traits that it compares with old images already stored in its database, all to an astonishing 83% accuracy.
Facial recognition has been receiving a lot of press lately. The news comes a week after several prominent privacy groups terminated negotiations with the U.S. government and the tech industry after failing to agree on even basic privacy protections around facial recognition. Moreover, dozens of churches worldwide are reportedly using facial recognition to track if its members are attending their services.
Here at the Lab, we have always been interested in the development of facial recognition technology, believing in its potential in enabling more personalized consumer experiences. But we also firmly believe that this technology should only be applied with clear consent from the consumers, preferably on an opt-in basis. Anything less would be a violation of consumer privacy, and should not be tolerated.
We’ve seen product recommendations highly targeted online, but not so offline. Glamour is trying to bring product recommendations OOH, piloting kiosks in subways which recommend products after detecting the colors of your outfit. While the targeting parameters are novel, it remains to be seen if shoppers open to purchases in these pop-up style environments.
Thanks to recently announced partnerships between Facebook and data providers Acxiom, Datalogix and Epsilon, marketers now have a greater degree of control than ever before when targeting their advertisements. These companies have compiled audience segments ready to be layered on top of the Facebook information-based targeting already available. These segments give marketers the ability to target based on the sites someone visits outside of Facebook, loyalty program memberships, household size, and more than 500 other criteria. With a billion-plus audience, Facebook has been attractive to marketers looking for broad spectrum coverage, and with these new additions, extremely detailed targeting is now a reality.
Facebook is expanding its custom audiences tools to let advertisers target based on information from third party sites like Datalogix, Epsilon, Acxiom, and BlueKai. This means a car company, for example, could target users on Facebook that have been ID-ed by third parties as being in the market to buy a new car. While privacy advocates will likely sound the emergency alarm, it’s an opportunity for big brands to increase efficiency and spend money on users likely to return the favor.
Facebook is testing out a new kind of ad targeting that matches offline purchases to Facebook profiles.
Prepare to be watched while you watch. Intel Media GM, Erik Huggers, has revealed that Intel will launch a set-top box this year to power an Internet television service including live television, on-demand, catch-up television, and apps. Hopes are high for the quality of the service, but concerns over privacy have already been voiced, as the box will feature a camera, which is rumored to be used to target advertisements to users. Advertisers have been excited for some time about this type of targeting technology, and this box could be a strong first step towards integrating it into home TV setups.
Gracenote is expecting to release a TV targeting platform later this year that would bring a level of precision comparable to online. The new service would support targeting based on gender, income and more by replacing ads run by broadcasters with those picked by Smart TV and set-top box providers. Additionally, there will be more transparent reporting associated as well.
Facebook Exchange: A New Way for Advertisers to Target Specific Users with Real-Time Bid Ads
Twitter rolling out ‘Tailored Trends’ now, tuned to your location and interests