DeepFashion
lastmod 2026-11-10created 2024-07-052 edits
DeepFashion is a large-scale clothing image dataset released by MMLab at the Chinese University of Hong Kong (CUHK). Hundreds of thousands of fashion photos are annotated with clothing categories, attributes, landmarks (keypoints on garments), and cross-item matching information, and it is used for training and evaluating models for clothing classification, attribute prediction, landmark detection, and product retrieval. There is also a follow-up version, DeepFashion2.