Cornell University researchers have found the motherlode of data to inform their studies about clothing trends around the world: Instagram. They've applied machine learning to a set of 100 million photos uploaded to the image sharing app, and while the results aren't earth shattering (red hats are big at Christmas!) they've paved the way for the anthropologists of the future.
The photos come from 44 cities across the world. A machine learning algorithm was trained to identify faces and articles of clothing. After weeding out photos without faces or a visible torso, the algorithm went to work on 15 million images. While findings were a bit basic this time around, it represents a step toward mining a massive data set that may help anthropologists conduct broad studies of culture and fashion in a way that's never been done before.
Read more about this research at MIT Technology Review.
. dpreview.com2017-6-16 00:30
