This AI system can give you fashion advice but bias likely to creep in

US-based researchers have developed a new artificial intelligence (AI) system that can give you fashion advice. Yes, you read that right. Researchers from the University of Texas have developed a new AI system named Fashion++ that can suggest tips to make certain outfits more fashionable by simply taking a look at their picture. According to researchers, this advanced AI system uses visual recognition systems to analyse the colour, pattern, texture and shape of garments in an image. The AI is so advanced that it also considers where edits will have the most impact. Based on that, it would offer alternative outfits to the user.

"We thought of it like a friend giving you feedback. It's also motivated by a practical idea: that we can work with a given outfit to make small changes so it's just a bit better," said Kristen Grauman, a professor at the University of Texas at Austin in the US.

READ | First 'AI' university launched in Abu Dhabi, could India be next?

How the AI was trained

Researchers have revealed Fashion++ was trained with the help of more than 10,000 images of outfits shared publicly on online websites, for fashion enthusiasts. Finding images of fashionable outfits was easy, said graduate student Kimberly Hsiao. However, what was more difficult was finding unfashionable images, which was proved a challenging task. Researchers mixed pictures of fashionable outfits to create less-fashionable examples and trained the system accordingly so that the AI could suggest what not to wear.

"As fashion styles evolve, the AI can continue to learn by giving it new images, which are abundant on the internet," Hsiao said.

READ | AI predicts effectiveness of antidepressants in patients: Study

Both, Grauman and Hsiao will showcase their system at the International Conference on Computer Vision in Seoul, South Korea next week. Researchers noted that like all AI systems, bias can creep in through the data sets for Fashion++. They pointed out that vintage looks are harder to recognise as stylish because training images came from the internet, which has been in wide use only since the 1990s.

Since user-submitted images were mostly from North America, styles from other parts of the world don't show up as much, researchers said. Another challenge is that many images of fashionable clothes appear on models, but bodies come in many sizes and shapes, affecting fashion choices, they said.

(With PTI inputs)