For iDesigner competition, we have sampled 50,000 runway images, spanning 50 fashion designers, from our large repository of proprietary, front row images from fashion shows over the past 15 years. The data includes a variety of fashion items: shoes, bags, dress, jackets etc. Each image in the training set has a single label, the designer's name, which you will try to predict for the test set.
The test set contains 5,000 similar images, of which ~30% will be used to generate the public leaderboard during the Kaggle competition.
For more details on iDesigner 2019 dataset & challenge, please see our Kaggle, Github, and FGVC 2019 page. The 2019 competition, sponsored by Hearst, is part of the FGVC6 workshop at CVPR.
Hearst Magazines Media, Inc. is the world's leading data-driven fashion media company. One of the largest publishers of monthly magazine media publishers, Hearst publishes 25 U.S. titles and over 300 international editions. Hearst's leadership in the fashion & luxury category is supported by brands such as Harper's BAZAAR, ELLE, Esquire and Marie Claire. Leveraging data from our 200M unique visitors, on-and-off site click stream data, and our brand advertising partners data, we're building a robust and disruptive capability to serve the fashion & luxury industry with insights to better build their businesses and grow market share.