Experimental app lets your phone judge your fashion sense
If you care about clothing styles in the first place, then it can be difficult, trying to determine if your outfit is fashionable. It would be great if your phone could tell you, and with the new Fashion++ app, it may soon be able to do just that.
The smartphone application was designed in a partnership between researchers from The University of Texas at Austin, Cornell Tech, Georgia Tech, and Facebook AI Research.
The artificial intelligence-based algorithms that it utilizes were "trained" using a database of over 10,000 images of outfits, which people had shared publicly on fashion websites. Among other things, those algorithms were able to visually recognize factors such as the color, pattern, texture and shape of garments.
The researchers also provided the system with examples of "what not to wear," by creating composite outfits consisting of clashing elements that were cut and pasted from multiple photos.
As a result, the app is now reportedly capable of assessing a smartphone photo of a user's outfit, then making suggestions such as selecting a longer jacket or a sleeveless top. That said, in its current form, it could still use some improvements.
For one thing, because its database was limited to clothing that was made after the internet went into widespread use, it doesn't recognize "retro" fashions. It's also mainly focused on North American clothing, along with garments worn by slim models. The researchers plan on addressing those problems as the technology is developed further.
"We are examining the interaction between how a person’s body is shaped and how the clothing would suit them," says U Texas' Prof. Kristen Grauman, who led the research along with grad student Kimberly Hsiao. "We’re excited to broaden the applicability to people of all body sizes and shapes by doing this research."