Matching Image and Sentence With Multi-Faceted Representations | IEEE Journals & Magazine | IEEE Xplore