Fast metric multi-view hashing for multimedia retrieval

Jian Zhu, Pengbo Hu, Bingqian Li, Yi Zhou

Abstract: Highlights•We propose a novel multi-view hash method, which achieves state-of-the-art results in multimedia retrieval.•We propose a deep metric loss to obtain information provided by dissimilar samples.•We use Context Gating to address the issue of inadequate fusion.•We employ the model compression technique to optimize the backbone networks.•We utilize a single deep metric loss to train the multi-view hash method.