Deep Uncoupled Discrete Hashing via Similarity Matrix Decomposition.

Dayan Wu,Qi Dai,Bo Li,Weiping Wang

DOI: https://doi.org/10.1145/3524021

2023-01-01

Abstract:Hashing has been drawing increasing attention in the task of large-scale image retrieval owing to its storage and computation efficiency, especially the recent asymmetric deep hashing methods. These approaches treat the query and database in an asymmetric way and can take full advantages of the whole training data. Though achieved the state-of-the-art performance, asymmetric deep hashing methods still suffer from the large quantization error and efficiency problem on large-scale dataset due to the tight coupling between query and database. In this paper, we propose a novel asymmetric hashing method, called D eep U ncoupled D iscrete H ashing (DUDH), for large-scale approximate nearest neighbor search. Instead of directly preserving the similarity between query and database, DUDH first exploits a small similarity-transfer image set to transfer the underlying semantic structures from database to query, and implicitly keep the desired similarity. As a result, the large similarity matrix is decomposed into two relatively small ones and query is decoupled from database. Then both database codes and similarity-transfer codes are directly learned during optimization. The quantization error of DUDH only exists in the process of preserving similarity between query and similarity-transfer set. By uncoupling query from database, the training cost of optimizing CNN model for query is no longer related to the size of database. Besides, to further accelerate the training process, we propose to optimize the similarity-transfer codes with a constant-approximation solution. In doing so, the training cost of optimizing similarity-transfer codes can be almost ignored. Extensive experiments on four widely used image retrieval benchmarks demonstrate that DUDH can achieve state-of-the-art retrieval performance with remarkable training cost reduction ( \(30\%-50\% \) relative).