From Selective Deep Convolutional Features to Compact Binary Representations for Image Retrieval



Do, Thanh-Toan ORCID: 0000-0002-6249-0848, Hoang, Tuan ORCID: 0000-0002-1076-8043, Tan, Dang-Khoa Le, Le, Huu, Nguyen, Tam V and Cheung, Ngai-Man
(2019) From Selective Deep Convolutional Features to Compact Binary Representations for Image Retrieval. ACM Transactions on Multimedia Computing, Communications, and Applications, 15 (2). 1 - 22.

[img] Text
1802.02899v3.pdf - Submitted Version

Download (4MB) | Preview

Abstract

<jats:p> In the large-scale image retrieval task, the two most important requirements are the discriminability of image representations and the efficiency in computation and storage of representations. Regarding the former requirement, Convolutional Neural Network is proven to be a very powerful tool to extract highly discriminative local descriptors for effective image search. Additionally, to further improve the discriminative power of the descriptors, recent works adopt fine-tuned strategies. In this article, taking a different approach, we propose a novel, computationally efficient, and competitive framework. Specifically, we first propose various strategies to compute masks, namely, <jats:bold> <jats:italic>SIFT-masks</jats:italic> </jats:bold> , <jats:bold> <jats:italic>SUM-mask</jats:italic> </jats:bold> , and <jats:bold> <jats:italic>MAX-mask</jats:italic> </jats:bold> , to select a representative subset of local convolutional features and eliminate redundant features. Our in-depth analyses demonstrate that proposed masking schemes are effective to address the burstiness drawback and improve retrieval accuracy. Second, we propose to employ recent embedding and aggregating methods that can significantly boost the feature discriminability. Regarding the computation and storage efficiency, we include a hashing module to produce very compact binary image representations. Extensive experiments on six image retrieval benchmarks demonstrate that our proposed framework achieves the state-of-the-art retrieval performances. </jats:p>

Item Type: Article
Depositing User: Symplectic Admin
Date Deposited: 18 Feb 2020 10:23
Last Modified: 15 Apr 2022 17:10
DOI: 10.1145/3314051
URI: https://livrepository.liverpool.ac.uk/id/eprint/3075400