Look More Fashionable On Instagram With Some Help From This Algorithm
Everyone wants to look their best in their Instagram pics. Let's be real — the whole point of taking and posting a selfie on the photo-sharing platform is to show the world how hot you are.
In order to make sure you look amazing in your Instagram photos, you might hold the camera a certain way, use lighting to your advantage or play around with filters until you find the one that will enhance your photo the most.
And then of course, what you wear plays a big part, too. Though you may get inspiration for your personal style from a variety of places – fashion blogs, celebrities and other Instagram users – there might actually be more science than art involved in picking an outfit to impress your followers.
Researchers from the University of Toronto and Spain's Institut de Robòtica i Informàtica Industrial created an algorithm that determines how fashionable people look in a photo and then suggests ways to improve their looks.
"Our aim here is to give a rich feedback to the user: not only whether the photograph is appealing or not, but also to make suggestions of what clothing or even the scenery the user could change in order to improve her/his look," the authors wrote in a paper detailing the algorithm at a recent computer vision and pattern recognition conference.
The researchers collected 144,169 user posts from the style social network Chictopia posted from March 2, 2008 to May 22, 2014, in which users showed off their outfits in photos. The fashionability of each post was based on public opinion, i.e. "votes," which are equivalent to "likes" on the photos, and the comments.
However, the algorithm that the researchers came up with takes several factors into account to determine how to make someone look more fashionable in a photo. These include the fashionability score, what the user is wearing, what the user looks like and the background and geographic setting of the photo.
Once an outfit is scored on fashionability, the algorithm then makes recommendations for how the user can increase that score, which you can see some examples of below.
The algorithm can also be used to find the "least fitting outfit," the best places to wear a particular outfit or which other types of users wear a certain ensemble the most.
The researchers also used this dataset spanning six years to analyze fashion trends over time in Manila in the Philippines and in Los Angeles. For instance, the popularity of high heels has skyrocketed over the past few years, while it has actually been on the decline among users in L.A. In that way, this algorithm could potentially be used by companies, fashion designers and tastemakers of all sorts to stay ahead of the curve and possibly determine what the next big trends will be.
Computer science. So hot right now.