Unsupervised Deep Hashing for Large-scale Visual Search

TL;DR

This paper introduces an unsupervised deep learning method for hashing that hierarchically transforms features into binary codes, improving large-scale visual search efficiency and accuracy.

Contribution

It proposes a novel deep hashing framework combining autoencoders and RBMs with constraints for better binary code learning.

Findings

Outperforms state-of-the-art hashing methods in visual search tasks.

Effectively models nonlinear feature-to-binary code mapping.

Reduces dimensionality in Hamming space with constraints.

Abstract

Learning based hashing plays a pivotal role in large-scale visual search. However, most existing hashing algorithms tend to learn shallow models that do not seek representative binary codes. In this paper, we propose a novel hashing approach based on unsupervised deep learning to hierarchically transform features into hash codes. Within the heterogeneous deep hashing framework, the autoencoder layers with specific constraints are considered to model the nonlinear mapping between features and binary codes. Then, a Restricted Boltzmann Machine (RBM) layer with constraints is utilized to reduce the dimension in the hamming space. Extensive experiments on the problem of visual search demonstrate the competitiveness of our proposed approach compared to state-of-the-art.