Powerful facial recognition algorithm used in Russian dating app
FindFace is a powerful Russian facial recognition app, developed by N-Tech Lab, that allows a mobile user to photograph strangers in a crowd and find their real identity by connecting them to their social media accounts with a purported 70 percent success rate.
According to The Guardian, FindFace uses image recognition technology to compare faces against profile pictures on the popular Russian social networking site, Vkontakte, that has over 200 million users.
FindFace is currently marketed as a dating app, but its founders are aiming to profit handsomely from the technology by licensing its algorithm to retail companies and law enforcement. The app has already been used by Russian police to find missing persons, identify arsonists in St. Petersburg, and solve cold cases using stills from old surveillance footage.
The app’s co-founders, Artem Kukharenko and Alexander Kabakov, claim their algorithm can search and cross-reference a billion photographs in a matter of seconds on a normal computer. Kaspersky Labs, an IT security firm, tested FindFace’s algorithm in April and found that the app is highly accurate. The app is so accurate that it has been selected by billionaire Mikhail Fridman for use to identify partygoers at his Alfa Future People electronic music festival.
The algorithm developed by N-Tech Lab, called FaceN, uses a “deep learning model” that recognizes facial features that are not perceptible to humans, in order to recognize faces in a faster, precise and more advanced way that people can.
The approach is so exact that FaceN claims to be able to identify a person in a “split second” and find their photograph among billions of other portraits in a central database. N-Tech Lab claims their system can be used on a “planetary scale” due to its efficient search algorithm, and due to their specially developed technology, which allows extracted feature data to be stored in a database in a compact fashion.
N-Tech Lab’s algorithm demonstrated its strong accuracy during the first phase of the MegaFace Challenge conducted by the University of Washington. As reported previously in BiometricUpdate.com, the MegaFace Challenge is the world’s first competition aimed at evaluating and improving the performance of face recognition algorithms on a “million person scale”. On a number of tests, the FaceN algorithm ranked number one in accuracy and compared highly to FaceNet, Google’s cloud-based facial recognition system.