A New Patch selection method based on Parsing and Saliency Detection for Person Re-identification

Yixiu Liu, Yunzhou Zhang, Sonya Coleman, Bir Bhanu, Shuangwei Liu

Research output: Contribution to journalArticle

Abstract

Person re-identification is an important technique towards automatic recognition of a person across non- overlapping cameras. In this paper, a novel patch selection method based on parsing and saliency de- tection is proposed. The algorithm is divided into two stages. The first stage, primary selection: Deep Decompositional Network (DNN) is adopted to parse a pedestrian image into semantic regions, then slid- ing window and color matching techniques are proposed to select pedestrian patches and remove back- ground patches. The second stage, secondary selection: saliency detection is utilized to select reliable patches according to saliency map. Finally, PHOG, HSV and SIFT features are extracted from these patches and fused with the global feature LOMO to compensate for the inherent errors of saliency detection. By applying the proposed method on such datasets as VIPeR, PRID2011, CUHK01, CUHK03, PRID 450S and iLIDS-VID, it is found that the proposed descriptor can produce results superior to many state-of-the-art feature representation methods for person identification.
LanguageEnglish
Pages86-99
Number of pages14
JournalNeurocomputing
Volume374
Early online date27 Sep 2019
DOIs
Publication statusPublished - 21 Jan 2020

Fingerprint

Color matching
Semantics
Cameras
Color
Pedestrians

Keywords

  • person re-identification
  • patch selection
  • pedestrian parsing
  • saliency detection
  • Feature fusion
  • Pedestrian parsing
  • Person re-identification
  • Patch selection
  • Saliency detection

Cite this

Liu, Yixiu ; Zhang, Yunzhou ; Coleman, Sonya ; Bhanu, Bir ; Liu, Shuangwei. / A New Patch selection method based on Parsing and Saliency Detection for Person Re-identification. In: Neurocomputing. 2020 ; Vol. 374. pp. 86-99.
@article{2e15e86b167542febdd83ebd459f8b34,
title = "A New Patch selection method based on Parsing and Saliency Detection for Person Re-identification",
abstract = "Person re-identification is an important technique towards automatic recognition of a person across non- overlapping cameras. In this paper, a novel patch selection method based on parsing and saliency de- tection is proposed. The algorithm is divided into two stages. The first stage, primary selection: Deep Decompositional Network (DNN) is adopted to parse a pedestrian image into semantic regions, then slid- ing window and color matching techniques are proposed to select pedestrian patches and remove back- ground patches. The second stage, secondary selection: saliency detection is utilized to select reliable patches according to saliency map. Finally, PHOG, HSV and SIFT features are extracted from these patches and fused with the global feature LOMO to compensate for the inherent errors of saliency detection. By applying the proposed method on such datasets as VIPeR, PRID2011, CUHK01, CUHK03, PRID 450S and iLIDS-VID, it is found that the proposed descriptor can produce results superior to many state-of-the-art feature representation methods for person identification.",
keywords = "person re-identification, patch selection, pedestrian parsing, saliency detection, Feature fusion, Pedestrian parsing, Person re-identification, Patch selection, Saliency detection",
author = "Yixiu Liu and Yunzhou Zhang and Sonya Coleman and Bir Bhanu and Shuangwei Liu",
year = "2020",
month = "1",
day = "21",
doi = "10.1016/j.neucom.2019.09.073",
language = "English",
volume = "374",
pages = "86--99",
journal = "Neurocomputing",
issn = "0925-2312",
publisher = "Elsevier",

}

A New Patch selection method based on Parsing and Saliency Detection for Person Re-identification. / Liu, Yixiu; Zhang, Yunzhou; Coleman, Sonya; Bhanu, Bir; Liu, Shuangwei.

In: Neurocomputing, Vol. 374, 21.01.2020, p. 86-99.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A New Patch selection method based on Parsing and Saliency Detection for Person Re-identification

AU - Liu, Yixiu

AU - Zhang, Yunzhou

AU - Coleman, Sonya

AU - Bhanu, Bir

AU - Liu, Shuangwei

PY - 2020/1/21

Y1 - 2020/1/21

N2 - Person re-identification is an important technique towards automatic recognition of a person across non- overlapping cameras. In this paper, a novel patch selection method based on parsing and saliency de- tection is proposed. The algorithm is divided into two stages. The first stage, primary selection: Deep Decompositional Network (DNN) is adopted to parse a pedestrian image into semantic regions, then slid- ing window and color matching techniques are proposed to select pedestrian patches and remove back- ground patches. The second stage, secondary selection: saliency detection is utilized to select reliable patches according to saliency map. Finally, PHOG, HSV and SIFT features are extracted from these patches and fused with the global feature LOMO to compensate for the inherent errors of saliency detection. By applying the proposed method on such datasets as VIPeR, PRID2011, CUHK01, CUHK03, PRID 450S and iLIDS-VID, it is found that the proposed descriptor can produce results superior to many state-of-the-art feature representation methods for person identification.

AB - Person re-identification is an important technique towards automatic recognition of a person across non- overlapping cameras. In this paper, a novel patch selection method based on parsing and saliency de- tection is proposed. The algorithm is divided into two stages. The first stage, primary selection: Deep Decompositional Network (DNN) is adopted to parse a pedestrian image into semantic regions, then slid- ing window and color matching techniques are proposed to select pedestrian patches and remove back- ground patches. The second stage, secondary selection: saliency detection is utilized to select reliable patches according to saliency map. Finally, PHOG, HSV and SIFT features are extracted from these patches and fused with the global feature LOMO to compensate for the inherent errors of saliency detection. By applying the proposed method on such datasets as VIPeR, PRID2011, CUHK01, CUHK03, PRID 450S and iLIDS-VID, it is found that the proposed descriptor can produce results superior to many state-of-the-art feature representation methods for person identification.

KW - person re-identification

KW - patch selection

KW - pedestrian parsing

KW - saliency detection

KW - Feature fusion

KW - Pedestrian parsing

KW - Person re-identification

KW - Patch selection

KW - Saliency detection

UR - http://www.scopus.com/inward/record.url?scp=85073013635&partnerID=8YFLogxK

U2 - 10.1016/j.neucom.2019.09.073

DO - 10.1016/j.neucom.2019.09.073

M3 - Article

VL - 374

SP - 86

EP - 99

JO - Neurocomputing

T2 - Neurocomputing

JF - Neurocomputing

SN - 0925-2312

ER -