It sounds like the premise of a dystopian sci-fi movie, but people in Russia are going nuts for an app that lets them take a photo of a stranger on the street, upload it to an app, and figure out their identity, with around 70 per cent reliability.
If that’s making you feel mighty nervous, it should — imagine sitting on a train and the guy opposite you is checking out your LinkedIn profile. The guy behind you has already found your Facebook page and is now cyberstalking your best friend. The girl next to you has found you on Tinder.
Creepy for a whole different reason is the way the app can also be used in reverse —upload a photo of your celebrity crush, and it will bring up a bunch of strangers nearby who look like them.
Meet FindFace, a face recognition app developed by Russian 20-somethings Artem Kukharenko and Alexander Kabakov. As ComputerWorld reports, the pair recently “upstaged” over 100 other facial recognition algorithms — including Google’s — in the University of Washington’s MegaFace challenge by achieving a 73.3 per cent accuracy on 1 million faces.
Since it launched two months ago in Russia, FindFace has already racked up 500,000 users and processed nearly 3 million searches,” according to Shaun Walker at The Guardian. It syncs up with Vkontakte — a popular social network in Russia and the former Soviet Union, which boasts more than 200 million accounts and nearly 1 billion photos.
“Three million searches in a database of nearly 1 billion photographs: that’s hundreds of trillions of comparisons, and all on four normal servers,” Kabakov told Walker. “With this algorithm, you can search through a billion photographs in less than a second from a normal computer.”
Basically, all you need to do is sneakily photograph someone on the street, upload the pic, and the app will serve up what it thinks is the most likely match, and 10 ‘bonus’ identities that look similar.
Source: www.sciencealert.com
