This Computer Program Can Spot Hipsters. Are you a biker, goth, raver, or hipster? Anew algorithm can tell 50% of the time, just from looking at your photos.
Fashion is a powerful tool. Within a split second, a person’s glasses, jeans, and haircut can communicate quite a bit about them–everything from age to socioeconomic status to connectedness to trends. Now, courtesy of scientists at the University of California, computers are getting a similar ability. Researchers there are developing a new algorithm that can distinguish someone’s urban tribe–whether he or she is a “biker, country, goth, heavy metal, hip hop, hipster, raver and surfer”–just by looking at a photos. And it’s accurate about 50% of the time. (Whereas a chance guess would only be right about 10% of the time.)
The software program keys in on personal details, like haircuts, hair color, makeup, jewelry, and tattoos, as well as colors and textures. It also takes this analysis a step further, by not only studying the key subject, but also every person photographed in any given scene. Each person essentially represents another datapoint about you.
See on ucsdnews.ucsd.edu