Over the last 10 years, the hipster archetype has been ever-changing: It’s been a vortex of style choices that include thick-rimmed glasses, fixed-gear bicycles, Pabst Blue Ribbon (slash Pilsner), mustaches, beards, denim vests, cold-pour coffee, hard lofts, five panel caps, Boston Terriers… and, well, the list goes on. Despite the complexities of the term, for the unscientific- minded, determining who’s a hipster—or more importantly, who isn’t one—is still pretty simple: You’ll know one when you see one.
Now, The Universty of California, San Diego (UCSD) Jacobs School of Engineering is developing a program that takes that concept seriously—it identifies hipsters on sight, by using an algorithm that scans photos. Developed by a group of students, the program is great at identifying subcultures immediately, and along with the dreaded H-word, it can also pick out goths, surfers, and bikers, and folks in formal wear, too.
As for the algorithm itself, Gizmag notes that the open-source program splits up a photo into six parts, analyzing the face, neck, head, torso, and arms. (We’re not sure what a goth neck would look like, aside from the fact that it’s likely pale.) Then, it crunches data on the type of clothing worn, the dominant hues present in the picture—meaning your favourite Instagram filter might actually say a lot about you—and the poses on hand.
“Image sharing via social networks has produced exciting opportunities for the computer vision community in areas including face, text, product and scene recognition,” the study states. “In this work we turn our attention to group photos of people at different social events.”
So, presumably, if the algorithm detects loads of camo and selvedge denim, it’s likely it’ll identify you as a hipster. If detects loads of black and high contrasts, you’re likely a goth. Sun-bleached hair paired with an unending tan? You’re a surfer. And if it detects whiteboy dreadlocks… Jesus, go get those cut. Immediately.
While the algorithm only predicts with a 48 per cent rate of accuracy—and it’s worth remembering that “urban tribes,” as its developers call it, are fluid in boundaries—its nefarious potential is already being explored: Image-reading algorithms mean that advertisers could target social groups with increasing accuracy, while Gizmag opines that the technology could be used to target political dissidents, too.
“Automatic recognition of such information in group photos offers the potential to improve recommendation services, context sensitive advertising and other social analysis applications,” the study’s introduction says. “Preliminary experimental results demonstrate our ability to categorize group photos in a socially meaningful manner.”
Translation: It identifies hipsters. As if we needed the help. Here’s a flow chart detailing how it works.