Your Face Is Big Data

This post is also available in: עברית (Hebrew)

Russian photography student, Yegor Tsvetkov, has created a demonstration that shows quite how easy it is for computers to identify individuals – complete strangers, even – in photos.

The St Petersburg-based photographer spent about six weeks shooting random people around his city’s streets and identifying them on the Russia’s most popular social network, Vkontakte, for the project Your Face Is Big Data. He did it using an app called FindFace that is based on algorithms used in neural networks.

“The idea of this project came to me when I first heard of the FindFace app,” said Tsvetkov, 21. “I instantly knew that I wanted to convey to people how this thing will work.”

Apparently, he was able to quite easily identify his subjects on the social network in about 70 percent of cases, even when there was a significant difference between the person’s presentation in real life and on the social platform. Some of the photos are from entirely different angles, with different lighting, and with obscured features.

This experiment demonstrates how far the technology has come that even using an online service, one is able to utilise the power of neural networks – the power wielded by more serious actors in the big data arena.

A simple app can find and compare photographs of reasonably good quality with those found on predefined online sources, but as we’ve seen at the latest iHLS Video Analytics conference, this is really small potatoes. The technology now exists to even track individuals across live video streams through the shape of their head, and we can only hope that authorities will use this technology for good.