How would you like it if you some stranger snapped your picture while you were out and about and then used that photo to find out your real name and other information about you?
FindFace facial recognition may not be brand new, but the app boasts of a 70% accuracy as in snap a photo of a stranger and then find out who that person is via their social media profile. For now, it works with VK, which is like a Russian version of Facebook, but it has been downloaded 500,000 times since February and has searched about three million photos.
“With this algorithm, you can search through a billion photographs in less than a second from a normal computer,” FindFace co-founder Alexander Kabakov told The Guardian. Talking about how the app could revolutionize dating, he added:
“If you see someone you like, you can photograph them, find their identity, and then send them a friend request. It also looks for similar people. So you could just upload a photo of a movie star you like, or your ex, and then find 10 girls who look similar to her and send them messages.”
If that’s not creepy enough for you, then there’s the Orwellian aspect of law enforcement and retail being the “real money-spinner” for the face recognition technology.
NTech Lab, which created FindFace, “upstaged” over 100 other facial recognition algorithms in the University of Washington’s MegaFace challenge by achieving a 73.3% accuracy on one million faces; it even beat Google’s “FaceNet” which had previously reached a 99.63% accuracy level (pdf) in different testing.
After winning the competition, NTech suggested its face recognition technology could be used by law enforcement “to search for and identify possible or wanted criminals using streaming video cameras,” or used by customs to identify people on watchlists, or to “verify and monitor” fans in real-time such as at sporting or entertainment events.
Since winning MegaFace, law enforcement has shown an interest. Russian police are using the technology in a pilot program via Moscow’s video camera network to search for criminals. NTech co-founder Alexander Kabakov told The Guardian, “It’s nuts: there were cases that had seen no movement for years, and now they are being solved.”
Sometimes it seems like face recognition is everywhere, used for social media sites and even at Church. The NSA has intercepted “millions” of photos daily and fed the “facial recognition quality images” into its facial recognition programs. The FBI was supposed to have 52 million photos in its NGI face recognition database by 2015 and was reportedly populating the database with photos snapped in the field. Face recognition is used by other law enforcement agencies such as police in Boston and San Diego.
The thing about FindFace is that it is not only fast, but it requires less computational resources than most facial recognition solutions; by lowering the computational costs, the Observer suggested, “N-Tech could make it feasible for the authorities to log everyone, just in case. Like the cops already do with license plates.”
Some people may find it really cool, but it’s already proved handy for stalking. Russian art school student Egor Tsvetkov took photos of strangers sitting across from him on the subway and then published his “Your face is big data” project; acting “like a Web stalker,” he linked the strangers’ photos to their VK profiles. Tsvetkov said that in theory, FindFace “could be used by a serial killer or a collector trying to hunt down a debtor.”
Immediately afterwards, the 2chan Russian imageboard launched a campaign using FindFace and VK to deanonymize and dox porn actresses. In the wrong hands, a tool like FindFace could be a death sentence for dissidents and victims of domestic abuse.
The idea of strangers or law enforcement using a face recognition app that requires little computational power to put a name to your face for no reason other than they want to? That should raise your privacy hackles. It may not be the only face recognition tech in the game, but it could spell an end to anonymity in public, open the door for harassment by strangers and put countless people at risk.
However, some people like Dr. Alec Couros, who have been an “indirect victim of catfishing,” would like for FindFace to be more widely available. He told The Observer, “If Facebook actually publicly used the tool, well … watch their user count drop like crazy.” Couros used FindFace to identify over 20 fake VK profiles using his pictures.
When Kasperksy Lab tested FindFace, it even identified a Kaspersky employee who doesn’t post any photos in any social networks; someone had posted his picture on a different social network and another user stole it and used it as his profile picture.
Kaspersky said FindFace “successfully found 9 of 10 test ‘victims’ in the office.” A positive note was that FindFace was two or three times less accurate if you take photos of strangers from a distance in the subway. If you crop and zoom, however, FindFace works again.
If a photo is taken during daylight, “with an average smartphone,” then it would be good enough to work for FindFace. If you don’t want to be “detected literally by any stranger with a phone,” then Kaspersky suggested deleting old photos from VK accounts, or trying to shield your face such as by wearing a hoodie.