Skip to Main Content
A customer has idea of different design elements associated with the garment, which he wants to buy. A questionnaire using items, which gives the representation of design elements, is prepared for fashion images. The representation of design elements is divided into several sections at different level. The relevancy between design elements and fashion image is determined by semantic network, in which an image object retrieval model is applied. The weights at each level are initialized and updated with relevance feedback technology. The experimental results show the efficiency of this method.