Image hashing is a popular technique applied to large scale content-basedvisual retrieval due to its compact and efficient binary codes. Our workproposes a new end-to-end deep network architecture for supervised hashingwhich directly learns binary codes from input images and maintains goodproperties over binary codes such as similarity preservation, independence, andbalancing. Furthermore, we also propose a new learning scheme that can copewith the binary constrained loss function. The proposed algorithm not only isscalable for learning over large-scale datasets but also outperformsstate-of-the-art supervised hashing methods, which are illustrated throughoutextensive experiments from various image retrieval benchmarks.
展开▼
