Visual Attributes for Fashion Analytics

In this chapter, we describe methods that leverage clothing and facial attributes as mid-level features for fashion recommendation and retrieval. We introduce a system called Magic Closet for recommending clothing for different occasions, and a system called Beauty E-Expert for hairstyle and facial makeup recommendation. For fashion retrieval, we describe a cross-domain clothing retrieval system, which receives as input a user photo of a particular clothing item taken in unconstrained conditions, and retrieves the exact same or similar item from online shopping catalogs. In each of these systems, we show the value of attribute-guided learning and describe approaches to transfer semantic concepts from large-scale uncluttered annotated data to challenging real-world imagery.