New software could help people decide what to wear for weddings, job interviews, or first dates.
Developed at the Chinese Academy of Science and the National University of Singapore, Magic Closet catalogs a user's wardrobe, using a Microsoft Kinect camera to identify key features such as sleeve length, color, and collar type. When asked for an outfit, the system finds the best matches from its database of 25,000 images of outfits skimmed from online shopping Web sites and the photo-sharing site Flickr. The database images were fed into Amazon's Mechanical Turk labor market, and the online workers categorized the outfits by attributes and determined whether the clothes were suitable for certain occasions. More than half of the workers had to agree for an attribute or occasion to be deemed appropriate.
Magic Closet also could be used by online shoppers to find clothes that go with items they already own. The team is adding personal taste to the model.
From New Scientist
View Full Article
No entries found