Structured deep hashing with convolutional neural networks for fast person re-identification

Lin Wu, Yang Wang, Zongyuan Ge, Qichang Hu, Xue Li

Research output: Contribution to journalArticleResearchpeer-review

21 Citations (Scopus)

Abstract

Given a pedestrian image as a query, the purpose of person re-identification is to identify the correct match from a large collection of gallery images depicting the same person captured by disjoint camera views. The critical challenge is how to construct a robust yet discriminative feature representation to capture the compounded variations in pedestrian appearance. To this end, deep learning methods have been proposed to extract hierarchical features against extreme variability of appearance. However, existing methods in this category generally neglect the efficiency in the matching stage whereas the searching speed of a re-identification system is crucial in real-world applications. In this paper, we present a novel deep hashing framework with Convolutional Neural Networks (CNNs) for fast person re-identification. Technically, we simultaneously learn both CNN features and hash functions to get robust yet discriminative features and similarity-preserving hash codes. Thereby, person re-identification can be resolved by efficiently computing and ranking the Hamming distances between images. A structured loss function defined over positive pairs and hard negatives is proposed to formulate a novel optimization problem so that fast convergence and more stable optimized solution can be attained. Extensive experiments on two benchmarks CUHK03 (Li et al., 2014) and Market-1501 (Zheng et al., 2015) show that the proposed deep architecture is efficacy over state-of-the-arts.

Original languageEnglish
Pages (from-to)63-73
Number of pages11
JournalComputer Vision and Image Understanding
Volume167
DOIs
Publication statusPublished - 1 Feb 2018
Externally publishedYes

Keywords

  • Convolutional neural networks
  • Deep hashing
  • Person re-identification
  • Structured embedding

Cite this

@article{0c83e34cbf9f493c82bed2f2d7d9138b,
title = "Structured deep hashing with convolutional neural networks for fast person re-identification",
abstract = "Given a pedestrian image as a query, the purpose of person re-identification is to identify the correct match from a large collection of gallery images depicting the same person captured by disjoint camera views. The critical challenge is how to construct a robust yet discriminative feature representation to capture the compounded variations in pedestrian appearance. To this end, deep learning methods have been proposed to extract hierarchical features against extreme variability of appearance. However, existing methods in this category generally neglect the efficiency in the matching stage whereas the searching speed of a re-identification system is crucial in real-world applications. In this paper, we present a novel deep hashing framework with Convolutional Neural Networks (CNNs) for fast person re-identification. Technically, we simultaneously learn both CNN features and hash functions to get robust yet discriminative features and similarity-preserving hash codes. Thereby, person re-identification can be resolved by efficiently computing and ranking the Hamming distances between images. A structured loss function defined over positive pairs and hard negatives is proposed to formulate a novel optimization problem so that fast convergence and more stable optimized solution can be attained. Extensive experiments on two benchmarks CUHK03 (Li et al., 2014) and Market-1501 (Zheng et al., 2015) show that the proposed deep architecture is efficacy over state-of-the-arts.",
keywords = "Convolutional neural networks, Deep hashing, Person re-identification, Structured embedding",
author = "Lin Wu and Yang Wang and Zongyuan Ge and Qichang Hu and Xue Li",
year = "2018",
month = "2",
day = "1",
doi = "10.1016/j.cviu.2017.11.009",
language = "English",
volume = "167",
pages = "63--73",
journal = "Computer Vision and Image Understanding",
issn = "1077-3142",
publisher = "Elsevier",

}

Structured deep hashing with convolutional neural networks for fast person re-identification. / Wu, Lin; Wang, Yang; Ge, Zongyuan; Hu, Qichang; Li, Xue.

In: Computer Vision and Image Understanding, Vol. 167, 01.02.2018, p. 63-73.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Structured deep hashing with convolutional neural networks for fast person re-identification

AU - Wu, Lin

AU - Wang, Yang

AU - Ge, Zongyuan

AU - Hu, Qichang

AU - Li, Xue

PY - 2018/2/1

Y1 - 2018/2/1

N2 - Given a pedestrian image as a query, the purpose of person re-identification is to identify the correct match from a large collection of gallery images depicting the same person captured by disjoint camera views. The critical challenge is how to construct a robust yet discriminative feature representation to capture the compounded variations in pedestrian appearance. To this end, deep learning methods have been proposed to extract hierarchical features against extreme variability of appearance. However, existing methods in this category generally neglect the efficiency in the matching stage whereas the searching speed of a re-identification system is crucial in real-world applications. In this paper, we present a novel deep hashing framework with Convolutional Neural Networks (CNNs) for fast person re-identification. Technically, we simultaneously learn both CNN features and hash functions to get robust yet discriminative features and similarity-preserving hash codes. Thereby, person re-identification can be resolved by efficiently computing and ranking the Hamming distances between images. A structured loss function defined over positive pairs and hard negatives is proposed to formulate a novel optimization problem so that fast convergence and more stable optimized solution can be attained. Extensive experiments on two benchmarks CUHK03 (Li et al., 2014) and Market-1501 (Zheng et al., 2015) show that the proposed deep architecture is efficacy over state-of-the-arts.

AB - Given a pedestrian image as a query, the purpose of person re-identification is to identify the correct match from a large collection of gallery images depicting the same person captured by disjoint camera views. The critical challenge is how to construct a robust yet discriminative feature representation to capture the compounded variations in pedestrian appearance. To this end, deep learning methods have been proposed to extract hierarchical features against extreme variability of appearance. However, existing methods in this category generally neglect the efficiency in the matching stage whereas the searching speed of a re-identification system is crucial in real-world applications. In this paper, we present a novel deep hashing framework with Convolutional Neural Networks (CNNs) for fast person re-identification. Technically, we simultaneously learn both CNN features and hash functions to get robust yet discriminative features and similarity-preserving hash codes. Thereby, person re-identification can be resolved by efficiently computing and ranking the Hamming distances between images. A structured loss function defined over positive pairs and hard negatives is proposed to formulate a novel optimization problem so that fast convergence and more stable optimized solution can be attained. Extensive experiments on two benchmarks CUHK03 (Li et al., 2014) and Market-1501 (Zheng et al., 2015) show that the proposed deep architecture is efficacy over state-of-the-arts.

KW - Convolutional neural networks

KW - Deep hashing

KW - Person re-identification

KW - Structured embedding

UR - http://www.scopus.com/inward/record.url?scp=85039054826&partnerID=8YFLogxK

U2 - 10.1016/j.cviu.2017.11.009

DO - 10.1016/j.cviu.2017.11.009

M3 - Article

VL - 167

SP - 63

EP - 73

JO - Computer Vision and Image Understanding

JF - Computer Vision and Image Understanding

SN - 1077-3142

ER -