Person Re-Identification by Common-Near-Neighbor Analysis

Wei LI; Masayuki MUKUNOKI; Yinghui KUANG; Yang WU; Michihiko MINOH

doi:10.1587/transinf.2014EDP7102

Person Re-Identification by Common-Near-Neighbor Analysis

Wei LI, Masayuki MUKUNOKI, Yinghui KUANG, Yang WU, Michihiko MINOH

Full Text Views

0

Share
Cite this

Summary :

Re-identifying the same person in different images is a distinct challenge for visual surveillance systems. Building an accurate correspondence between highly variable images requires a suitable dissimilarity measure. To date, most existing measures have used adapted distance based on a learned metric. Unfortunately, real-world human image data, which tends to show large intra-class variations and small inter-class differences, continues to prevent these measures from achieving satisfactory re-identification performance. Recognizing neighboring distribution can provide additional useful information to help tackle the deviation of the to-be-measured samples, we propose a novel dissimilarity measure from the neighborhood-wise relative information perspective, which can deliver the effectiveness of those well-distributed samples to the badly-distributed samples to make intra-class dissimilarities smaller than inter-class dissimilarities, in a learned discriminative space. The effectiveness of this method is demonstrated by explanation and experimentation.

Publication: IEICE TRANSACTIONS on Information Vol.E97-D No.11 pp.2935-2946

Publication Date: 2014/11/01

Publicized

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2014EDP7102

Type of Manuscript: PAPER

Category: Image Recognition, Computer Vision

Authors

Wei LI
  Kyoto University
Masayuki MUKUNOKI
  Kyoto University
Yinghui KUANG
  Southeast University
Yang WU
  Kyoto University
Michihiko MINOH
  Kyoto University

Keyword

person re-identification, common-near-neighbor analysis, learned metric, visual surveillance

Cite this

Copy

Wei LI, Masayuki MUKUNOKI, Yinghui KUANG, Yang WU, Michihiko MINOH, "Person Re-Identification by Common-Near-Neighbor Analysis" in IEICE TRANSACTIONS on Information, vol. E97-D, no. 11, pp. 2935-2946, November 2014, doi: 10.1587/transinf.2014EDP7102.
Abstract: Re-identifying the same person in different images is a distinct challenge for visual surveillance systems. Building an accurate correspondence between highly variable images requires a suitable dissimilarity measure. To date, most existing measures have used adapted distance based on a learned metric. Unfortunately, real-world human image data, which tends to show large intra-class variations and small inter-class differences, continues to prevent these measures from achieving satisfactory re-identification performance. Recognizing neighboring distribution can provide additional useful information to help tackle the deviation of the to-be-measured samples, we propose a novel dissimilarity measure from the neighborhood-wise relative information perspective, which can deliver the effectiveness of those well-distributed samples to the badly-distributed samples to make intra-class dissimilarities smaller than inter-class dissimilarities, in a learned discriminative space. The effectiveness of this method is demonstrated by explanation and experimentation.
URL: https://globals.ieice.org/en_transactions/information/10.1587/transinf.2014EDP7102/_p

Copy

@ARTICLE{e97-d_11_2935,
author={Wei LI, Masayuki MUKUNOKI, Yinghui KUANG, Yang WU, Michihiko MINOH, },
journal={IEICE TRANSACTIONS on Information},
title={Person Re-Identification by Common-Near-Neighbor Analysis},
year={2014},
volume={E97-D},
number={11},
pages={2935-2946},
abstract={Re-identifying the same person in different images is a distinct challenge for visual surveillance systems. Building an accurate correspondence between highly variable images requires a suitable dissimilarity measure. To date, most existing measures have used adapted distance based on a learned metric. Unfortunately, real-world human image data, which tends to show large intra-class variations and small inter-class differences, continues to prevent these measures from achieving satisfactory re-identification performance. Recognizing neighboring distribution can provide additional useful information to help tackle the deviation of the to-be-measured samples, we propose a novel dissimilarity measure from the neighborhood-wise relative information perspective, which can deliver the effectiveness of those well-distributed samples to the badly-distributed samples to make intra-class dissimilarities smaller than inter-class dissimilarities, in a learned discriminative space. The effectiveness of this method is demonstrated by explanation and experimentation.},
keywords={},
doi={10.1587/transinf.2014EDP7102},
ISSN={1745-1361},
month={November},}

Copy

TY - JOUR
TI - Person Re-Identification by Common-Near-Neighbor Analysis
T2 - IEICE TRANSACTIONS on Information
SP - 2935
EP - 2946
AU - Wei LI
AU - Masayuki MUKUNOKI
AU - Yinghui KUANG
AU - Yang WU
AU - Michihiko MINOH
PY - 2014
DO - 10.1587/transinf.2014EDP7102
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E97-D
IS - 11
JA - IEICE TRANSACTIONS on Information
Y1 - November 2014
AB - Re-identifying the same person in different images is a distinct challenge for visual surveillance systems. Building an accurate correspondence between highly variable images requires a suitable dissimilarity measure. To date, most existing measures have used adapted distance based on a learned metric. Unfortunately, real-world human image data, which tends to show large intra-class variations and small inter-class differences, continues to prevent these measures from achieving satisfactory re-identification performance. Recognizing neighboring distribution can provide additional useful information to help tackle the deviation of the to-be-measured samples, we propose a novel dissimilarity measure from the neighborhood-wise relative information perspective, which can deliver the effectiveness of those well-distributed samples to the badly-distributed samples to make intra-class dissimilarities smaller than inter-class dissimilarities, in a learned discriminative space. The effectiveness of this method is demonstrated by explanation and experimentation.
ER -