Unsupervised local regressive attributes for pedestrian re-identification

Billy Peralta, Luis Caro, Alvaro Soto

Resultado de la investigación: Conference contribution

Resumen

Discovering of attributes is a challenging task in computer vision due to uncertainty about the attributes, which is caused mainly by the lack of semantic meaning in image parts. A usual scheme for facing attribute discovering is to divide the feature space using binary variables. Moreover, we can assume to know the attributes and by using expert information we can give a degree of attribute beyond only two values. Nonetheless, a binary variable could not be very informative, and we could not have access to expert information. In this work, we propose to discover linear regressive codes using image regions guided by a supervised criteria where the obtained codes obtain better generalization properties. We found that the discovered regressive codes can be successfully re-used in other visual datasets. As a future work, we plan to explore richer codification structures than lineal mapping considering efficient computation.

Idioma originalEnglish
Título de la publicación alojadaProgress in Pattern Recognition, Image Analysis, Computer Vision, and Applications - 22nd Iberoamerican Congress, CIARP 2017, Proceedings
EditoresSergio Velastin, Marcelo Mendoza
EditorialSpringer Verlag
Páginas517-524
Número de páginas8
ISBN (versión impresa)9783319751924
DOI
EstadoPublished - 1 ene 2018
Evento22nd Iberoamerican Congress on Pattern Recognition, CIARP 2017 - Valparaiso, Chile
Duración: 7 nov 201710 nov 2017

Serie de la publicación

NombreLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volumen10657 LNCS
ISSN (versión impresa)0302-9743
ISSN (versión digital)1611-3349

Conference

Conference22nd Iberoamerican Congress on Pattern Recognition, CIARP 2017
PaísChile
CiudadValparaiso
Período7/11/1710/11/17

Huella dactilar

Computer vision
Semantics
Attribute
Binary Variables
Feature Space
Computer Vision
Divides
Uncertainty

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Citar esto

Peralta, B., Caro, L., & Soto, A. (2018). Unsupervised local regressive attributes for pedestrian re-identification. En S. Velastin, & M. Mendoza (Eds.), Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications - 22nd Iberoamerican Congress, CIARP 2017, Proceedings (pp. 517-524). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10657 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-75193-1_62
Peralta, Billy ; Caro, Luis ; Soto, Alvaro. / Unsupervised local regressive attributes for pedestrian re-identification. Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications - 22nd Iberoamerican Congress, CIARP 2017, Proceedings. editor / Sergio Velastin ; Marcelo Mendoza. Springer Verlag, 2018. pp. 517-524 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{ec8b240549474cd097f716c07bb53ddd,
title = "Unsupervised local regressive attributes for pedestrian re-identification",
abstract = "Discovering of attributes is a challenging task in computer vision due to uncertainty about the attributes, which is caused mainly by the lack of semantic meaning in image parts. A usual scheme for facing attribute discovering is to divide the feature space using binary variables. Moreover, we can assume to know the attributes and by using expert information we can give a degree of attribute beyond only two values. Nonetheless, a binary variable could not be very informative, and we could not have access to expert information. In this work, we propose to discover linear regressive codes using image regions guided by a supervised criteria where the obtained codes obtain better generalization properties. We found that the discovered regressive codes can be successfully re-used in other visual datasets. As a future work, we plan to explore richer codification structures than lineal mapping considering efficient computation.",
keywords = "Attribute discovery, Pedestrian re-identification, Unsupervised learning",
author = "Billy Peralta and Luis Caro and Alvaro Soto",
year = "2018",
month = "1",
day = "1",
doi = "10.1007/978-3-319-75193-1_62",
language = "English",
isbn = "9783319751924",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "517--524",
editor = "Sergio Velastin and Marcelo Mendoza",
booktitle = "Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications - 22nd Iberoamerican Congress, CIARP 2017, Proceedings",
address = "Germany",

}

Peralta, B, Caro, L & Soto, A 2018, Unsupervised local regressive attributes for pedestrian re-identification. En S Velastin & M Mendoza (eds.), Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications - 22nd Iberoamerican Congress, CIARP 2017, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10657 LNCS, Springer Verlag, pp. 517-524, 22nd Iberoamerican Congress on Pattern Recognition, CIARP 2017, Valparaiso, Chile, 7/11/17. https://doi.org/10.1007/978-3-319-75193-1_62

Unsupervised local regressive attributes for pedestrian re-identification. / Peralta, Billy; Caro, Luis; Soto, Alvaro.

Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications - 22nd Iberoamerican Congress, CIARP 2017, Proceedings. ed. / Sergio Velastin; Marcelo Mendoza. Springer Verlag, 2018. p. 517-524 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10657 LNCS).

Resultado de la investigación: Conference contribution

TY - GEN

T1 - Unsupervised local regressive attributes for pedestrian re-identification

AU - Peralta, Billy

AU - Caro, Luis

AU - Soto, Alvaro

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Discovering of attributes is a challenging task in computer vision due to uncertainty about the attributes, which is caused mainly by the lack of semantic meaning in image parts. A usual scheme for facing attribute discovering is to divide the feature space using binary variables. Moreover, we can assume to know the attributes and by using expert information we can give a degree of attribute beyond only two values. Nonetheless, a binary variable could not be very informative, and we could not have access to expert information. In this work, we propose to discover linear regressive codes using image regions guided by a supervised criteria where the obtained codes obtain better generalization properties. We found that the discovered regressive codes can be successfully re-used in other visual datasets. As a future work, we plan to explore richer codification structures than lineal mapping considering efficient computation.

AB - Discovering of attributes is a challenging task in computer vision due to uncertainty about the attributes, which is caused mainly by the lack of semantic meaning in image parts. A usual scheme for facing attribute discovering is to divide the feature space using binary variables. Moreover, we can assume to know the attributes and by using expert information we can give a degree of attribute beyond only two values. Nonetheless, a binary variable could not be very informative, and we could not have access to expert information. In this work, we propose to discover linear regressive codes using image regions guided by a supervised criteria where the obtained codes obtain better generalization properties. We found that the discovered regressive codes can be successfully re-used in other visual datasets. As a future work, we plan to explore richer codification structures than lineal mapping considering efficient computation.

KW - Attribute discovery

KW - Pedestrian re-identification

KW - Unsupervised learning

UR - http://www.scopus.com/inward/record.url?scp=85042213112&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-75193-1_62

DO - 10.1007/978-3-319-75193-1_62

M3 - Conference contribution

AN - SCOPUS:85042213112

SN - 9783319751924

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 517

EP - 524

BT - Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications - 22nd Iberoamerican Congress, CIARP 2017, Proceedings

A2 - Velastin, Sergio

A2 - Mendoza, Marcelo

PB - Springer Verlag

ER -

Peralta B, Caro L, Soto A. Unsupervised local regressive attributes for pedestrian re-identification. En Velastin S, Mendoza M, editores, Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications - 22nd Iberoamerican Congress, CIARP 2017, Proceedings. Springer Verlag. 2018. p. 517-524. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-75193-1_62