Just when you think researchers have found the wackiest possible use for a neural net yet, another team finds an even more novel use for artificial intelligence. Take Facebook’s new Fashion++ AI. It’s a program that will help you become a fashionista.
Facebook is embarking on a practical use case for AI that could help people become more fashionable. Its latest experiment called Fashion++ uses a deep image-generation neural network to suggest ways that people can improve their outfits by adding or swapping items. It will even offer suggestions like tucking in a shirt or rolling up sleeves to tweak existing clothing.
Focusing on practicality, the algorithm is aimed at making small edits as opposed to recommending a whole new outfit. To train the classifier, thousands of fashionable images were inputted to serve as the ground truth. Then, the least likely counterparts to those outfits were bootstrapped to determine what’s unfashionable.
To teach the algorithm what constitutes good style, researchers used more than 10,000 user-submitted street style images from Chictopia. The software can identify details including silhouette, texture, shape, colour and pattern. The researchers then created “unfashionable” looks by digitally replacing one piece of an outfit with the least similar garment. These inputs were used to develop a “fashionability classifier” — essentially a scale that scores a look on stylishness.
When a user submits a photo, Fashion++ then proposes changes that would raise its score. Suggestions are shown using digital renders drawn from Facebook’s inventory of clothing images. To support the software’s recommendations, 300 crowdsourced workers from Amazon Mechanical Turk rated its advice. The researchers say that the evaluators not only agreed with Fashion++’s recommendations but also found them useful and actionable.
This experiment if realized would mark a significant breakthrough Artificial Intelligence as it would give it a more creative and assistive role. Facebook claims that the system already works, stating “human evaluators find the Fashion++ suggestions not only fashionable but also easy to implement.”