Dense appearance modeling and efficient learning of camera transitions for person re-identification

Martin Hirzer, Csaba Beleznai, Martin Köstinger, Peter Roth, Horst Bischof

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

Abstract

One central task in many visual surveillance scenarios is person re-identification, i.e., recognizing an individual person across a network of spatially disjoint cameras. Most successful recognition approaches are either based on direct modeling of the human appearance or on machine learning. In this work, we aim at taking advantage of both directions of research. On the one hand side, we compute a descriptive appearance representation encoding the vertical color structure of pedestrians. To improve the classification results, we additionally estimate the transition between two cameras using a pair-wisely estimated metric. In particular, we introduce 4D spatial color histograms and adopt Large Margin Nearest Neighbor (LMNN) metric learning. The approach is demonstrated for two publicly available datasets, showing competitive results, however, on lower computational costs.
Original languageEnglish
Title of host publicationProceedings of the IEEE International Conference on Image Processing (ICIP)
Publisher.
Pages1617-1620
DOIs
Publication statusPublished - 2012
EventInternational Conference on Image Processing - Orlando, United States
Duration: 30 Sep 20123 Oct 2012

Conference

ConferenceInternational Conference on Image Processing
CountryUnited States
CityOrlando
Period30/09/123/10/12

Fingerprint

Cameras
Color
Learning systems
Costs

Fields of Expertise

  • Information, Communication & Computing

Cite this

Hirzer, M., Beleznai, C., Köstinger, M., Roth, P., & Bischof, H. (2012). Dense appearance modeling and efficient learning of camera transitions for person re-identification. In Proceedings of the IEEE International Conference on Image Processing (ICIP) (pp. 1617-1620). .. https://doi.org/10.1109/ICIP.2012.6467185

Dense appearance modeling and efficient learning of camera transitions for person re-identification. / Hirzer, Martin; Beleznai, Csaba; Köstinger, Martin; Roth, Peter; Bischof, Horst.

Proceedings of the IEEE International Conference on Image Processing (ICIP). ., 2012. p. 1617-1620.

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

Hirzer, M, Beleznai, C, Köstinger, M, Roth, P & Bischof, H 2012, Dense appearance modeling and efficient learning of camera transitions for person re-identification. in Proceedings of the IEEE International Conference on Image Processing (ICIP). ., pp. 1617-1620, International Conference on Image Processing, Orlando, United States, 30/09/12. https://doi.org/10.1109/ICIP.2012.6467185
Hirzer M, Beleznai C, Köstinger M, Roth P, Bischof H. Dense appearance modeling and efficient learning of camera transitions for person re-identification. In Proceedings of the IEEE International Conference on Image Processing (ICIP). . 2012. p. 1617-1620 https://doi.org/10.1109/ICIP.2012.6467185
Hirzer, Martin ; Beleznai, Csaba ; Köstinger, Martin ; Roth, Peter ; Bischof, Horst. / Dense appearance modeling and efficient learning of camera transitions for person re-identification. Proceedings of the IEEE International Conference on Image Processing (ICIP). ., 2012. pp. 1617-1620
@inproceedings{f68f5a4947c94ce9809f0ca0e8ff1dd1,
title = "Dense appearance modeling and efficient learning of camera transitions for person re-identification",
abstract = "One central task in many visual surveillance scenarios is person re-identification, i.e., recognizing an individual person across a network of spatially disjoint cameras. Most successful recognition approaches are either based on direct modeling of the human appearance or on machine learning. In this work, we aim at taking advantage of both directions of research. On the one hand side, we compute a descriptive appearance representation encoding the vertical color structure of pedestrians. To improve the classification results, we additionally estimate the transition between two cameras using a pair-wisely estimated metric. In particular, we introduce 4D spatial color histograms and adopt Large Margin Nearest Neighbor (LMNN) metric learning. The approach is demonstrated for two publicly available datasets, showing competitive results, however, on lower computational costs.",
author = "Martin Hirzer and Csaba Beleznai and Martin K{\"o}stinger and Peter Roth and Horst Bischof",
year = "2012",
doi = "http://dx.doi.org/10.1109/ICIP.2012.6467185",
language = "English",
pages = "1617--1620",
booktitle = "Proceedings of the IEEE International Conference on Image Processing (ICIP)",
publisher = ".",

}

TY - GEN

T1 - Dense appearance modeling and efficient learning of camera transitions for person re-identification

AU - Hirzer, Martin

AU - Beleznai, Csaba

AU - Köstinger, Martin

AU - Roth, Peter

AU - Bischof, Horst

PY - 2012

Y1 - 2012

N2 - One central task in many visual surveillance scenarios is person re-identification, i.e., recognizing an individual person across a network of spatially disjoint cameras. Most successful recognition approaches are either based on direct modeling of the human appearance or on machine learning. In this work, we aim at taking advantage of both directions of research. On the one hand side, we compute a descriptive appearance representation encoding the vertical color structure of pedestrians. To improve the classification results, we additionally estimate the transition between two cameras using a pair-wisely estimated metric. In particular, we introduce 4D spatial color histograms and adopt Large Margin Nearest Neighbor (LMNN) metric learning. The approach is demonstrated for two publicly available datasets, showing competitive results, however, on lower computational costs.

AB - One central task in many visual surveillance scenarios is person re-identification, i.e., recognizing an individual person across a network of spatially disjoint cameras. Most successful recognition approaches are either based on direct modeling of the human appearance or on machine learning. In this work, we aim at taking advantage of both directions of research. On the one hand side, we compute a descriptive appearance representation encoding the vertical color structure of pedestrians. To improve the classification results, we additionally estimate the transition between two cameras using a pair-wisely estimated metric. In particular, we introduce 4D spatial color histograms and adopt Large Margin Nearest Neighbor (LMNN) metric learning. The approach is demonstrated for two publicly available datasets, showing competitive results, however, on lower computational costs.

U2 - http://dx.doi.org/10.1109/ICIP.2012.6467185

DO - http://dx.doi.org/10.1109/ICIP.2012.6467185

M3 - Conference contribution

SP - 1617

EP - 1620

BT - Proceedings of the IEEE International Conference on Image Processing (ICIP)

PB - .

ER -