Is it just for clothes? How do you get the results?
We use image recognition and scrape items from social media, partner databases and crowdsourcing.
Also testing with Beta users to snap what they're wearing and let them tag semantically ("night out","office","summer wear", "goth").