Subway photographer connects random photos to people’s social media profiles

Егор Цветков (Egor Tsvetkov), a photographer in Russia, has taken photos of random people on the subway and connected them to social media portraits and complete profiles using face matching technology. This is a game changer.

It used to be that technology was good enough to say whether two photos appeared to be of the same person. We’ve now reached an inflection point where one input photo can (mostly) be used to find the matching person among tens of millions of people, and where the processing power used is low enough for that service to be free. This is a complete game changer.

The gravity of this doesn’t really hit you until you see the examples, where the photos are taken under radically different lighting and angles than the portrait photos, and sometimes with different facial hair, too. What’s more, this photographer used a freely available photo matching service – FindFace.ru – which has already imported a vast amount of (all?) photos on vKontakte, which is Russia’s equivalent of Facebook, and let a neural network study all of those photos.

As a result, this photographer claims a 70% success rate of matching random people photographed on the subway, in completely different lighting conditions, with their complete social media profile.

You cannot hold back the mere existence of technology. In less than five years, there will be CCTV cameras which list all the people being currently in-frame with their name and social media portrait. Shortly thereafter, law enforcement will use this for automated warrant spotting, RoboCop style – or will at least very much want to.

In ten years, the service will be available for personal use – superimposing people’s social portraits and names directly onto your field of view, using glasses or the projection contacts already in prototype.

And in maybe fifteen years, marketers will scream and cry to have the ability to provide “targeted messaging”, of the kind we saw in Minority Report – where billboards change in front of you to match your particular interests, or what the marketers think are your interests.

With the technology available as a free service, scaling up the processing power is just a matter of Moore’s Law and throwing money at it. Who’s going to provide the neural network and the photo databases? There’s obviously giants like Facebook and Google, but this would also a new potential monetization for any service where a lot of people have sent photos – like Tinder. Real-time photo matching for surveillance and convenience use.

Privacy remains your own responsibility.

UPDATE: Also see the Hacker News discussion on the subject.