Publication

ZOGHBI, Susana, HEYMAN, Geert, GOMEZ, Juan-Carlos & MOENS, Marie-Francine (2016) Cross-modal Fashion Search. In Proceedings of the 22nd International Conference on MultiMedia Modelling (Lecture Notes in Computer Science 9517) (pp. 367-373).

Download

View software demo

In this demo we show our latest developments in multimodal (visual-textual) search within the fashion domain. Our algorithms allow machines to perform two challenging tasks. First, given a textual query without any visual information, our system finds images that display the visual properties in the query. Secondly, given a query image without any supplementing text, our system generates text that describes the visual properties in the image. Remarkably, images shown in the results below correspond to items that our algorithm has never seen before. The retrieval models were made by Susana Zoghbi, Geert Heyman and Juan Carlos. The online demo was developed by Anirudh Tomer.

License