Cross-Modal Hashing

Cross-modal retrieval [1] aims to retrieve semantically relevant items in other modalities given queries from a specific modality. To support large-scale cross-modal retrieval, hashing [2] has been proposed due to its low storage cost and high retrieval speed. The main idea of cross-modal hashing is to encode the heterogeneous multi-modal data into a common isomorphic Hamming space while preserving their semantic similarities.