This dataset contains annotator judgments of the accuracy and reasonableness of a set of apparel products as responses to queries, where each query consists of a product together with a short caption describing how the response product should be different. There are three categories of products: 134 “dresses”, 100 “gowns” and 100 “sundresses”, each represented by an image to which the dataset provides a URL. Each of 40 queries (20 dresses, 10 gowns, 10 sundresses) consists of one of the products plus a short caption.

Dataset has been added to your cart