Dividing people into popular subcultures may be the quickest way to determine whether you have anything in common with someone but it wasn’t the kind of thing you would expect scientists to take a particular interest in until now.
Scientists from the University of California, San Diego have developed algorithms to determine an individual’s ‘urban tribe’ or subculture by analysing photos. There’s 11 categories used including ‘Surfer’, ‘Punk’, ‘Goth’, ‘Biker’ and ‘Hipster’. It comes as no surprise that this technology is part of a greater scheme to use social media to target specific demographics more effectively, using the abundance of free information we supply daily to plan marketing strategies, create applications and place products exactly where they’re most effective. The study was presented at the British Machine Vision Conference in September and is said to ‘present a compelling opportunity to analyse the social identity of individuals captured within (online) images’ according to Wired.com.
The algorithms work by dividing human characteristics into six main features – the head, top of the head, face, neck, torso and arms, using these to dig deeper into specifics like hair colour, hair style, accessories, tattoos, make-up and even textures. In many cases this would place a lot of wearers of oversized, thick-rimmed glasses neatly into the ‘Hipster’ category, along with half-shaven heads and checked shirt wearers. However the researchers admit there’s still some way to go to achieve a high level of accuracy, for example, your long, dark-haired aunt with porcelain skin and black jumper may easily end up categorised as a goth on the basis of her choice of clothing colour, complexion and straightened hair.
As it stands, the algorithms are correct at least 48% of the time, with more work needed in order to put their accuracy on par with human identification and make the technology more meaningful to marketers and social scientists alike.