Aimee Song has 2.2 million followers on Instagram. Photo: @songofstyle/Instagram
Scientists might hold the key to collecting more Instagram likes for an #OOTD post. Four researchers from the University of Toronto devised an algorithm to analyze photos, rank them, and even offer suggestions on how to improve an outfit (overwhelmingly: wear heels).
“People want to look good; business or casual, elegant or sporty, sexy but not slutty, and of course trendy, particularly so when putting their picture online,” they wrote, as part of their reasoning for conducting the study, which was published on The Computer Vision.
Using math much more complicated than what I was ever taught – thanks college! — the team analyzed 144,169 posts on chictopia.com, an aggregate site of people posting images of their looks with user-enabled voting. Based on the amount of votes received (similar to likes on Facebook and Instagram), positive and negative comments, tags, and other factors, data’s compiled into a fashionability factor.
This ranking’s based on clothing, but also setting, how the photo was taken, its composition, if the subject’s visually appealing, their age, and more. Additionally, just because a particular garment’s considered chic, doesn’t make the post so because “people typically also judge how well the garments align with someone’s “look,” body characteristics, or even personality.”
While the individual benefits of the algorithm are obvious (and wonderful), it also has the ability to predict trends down to a specific location. The researchers believe that their development is “an important first step to be able to build more complex and powerful models that will be able to understand fashion, trends” and even “improve the experience of users in the modern day society.”
Because what society really needs is science informing bloggers how to dress better.