Strike a pose: Tracking people by finding stylized poses

Deva Ramanan, D. A. Forsyth, Andrew Zisserman

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

We develop an algorithm for finding and kinematically tracking multiple people in long sequences. Our basic assumption is that people tend to take on certain canonical poses, even when performing unusual activities like throwing a baseball or figure skating. We build a person detector that quite accurately detects and localizes limbs of people in lateral walking poses. We use the estimated limbs from a detection to build a discriminative appearance model; we assume the features that discriminate a figure in one frame will discriminate the figure in other frames. We then use the models as limb detectors in a pictorial structure framework, detecting figures in unrestricted poses in both previous and successive frames. We have run our tracker on hundreds of thousands of frames, and present and apply a methodology for evaluating tracking on such a large scale. We test our tracker on real sequences including a feature-length film, an hour of footage from a public park, and various sports sequences. We find that we can quite accurately automatically find and track multiple people interacting with each other while performing fast and unusual motions.

Original languageEnglish (US)
Title of host publicationProceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005
PublisherIEEE Computer Society
Pages271-278
Number of pages8
ISBN (Print)0769523722, 9780769523729
DOIs
StatePublished - Jan 1 2005
Externally publishedYes
Event2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005 - San Diego, CA, United States
Duration: Jun 20 2005Jun 25 2005

Publication series

NameProceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005
VolumeI

Other

Other2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005
CountryUnited States
CitySan Diego, CA
Period6/20/056/25/05

Fingerprint

Detectors
Sports

ASJC Scopus subject areas

  • Engineering(all)

Cite this

Ramanan, D., Forsyth, D. A., & Zisserman, A. (2005). Strike a pose: Tracking people by finding stylized poses. In Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005 (pp. 271-278). [1467278] (Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005; Vol. I). IEEE Computer Society. https://doi.org/10.1109/CVPR.2005.335

Strike a pose : Tracking people by finding stylized poses. / Ramanan, Deva; Forsyth, D. A.; Zisserman, Andrew.

Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005. IEEE Computer Society, 2005. p. 271-278 1467278 (Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005; Vol. I).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Ramanan, D, Forsyth, DA & Zisserman, A 2005, Strike a pose: Tracking people by finding stylized poses. in Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005., 1467278, Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005, vol. I, IEEE Computer Society, pp. 271-278, 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005, San Diego, CA, United States, 6/20/05. https://doi.org/10.1109/CVPR.2005.335
Ramanan D, Forsyth DA, Zisserman A. Strike a pose: Tracking people by finding stylized poses. In Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005. IEEE Computer Society. 2005. p. 271-278. 1467278. (Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005). https://doi.org/10.1109/CVPR.2005.335
Ramanan, Deva ; Forsyth, D. A. ; Zisserman, Andrew. / Strike a pose : Tracking people by finding stylized poses. Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005. IEEE Computer Society, 2005. pp. 271-278 (Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005).
@inproceedings{d03ac497abcd4a43859568ce7bd871cf,
title = "Strike a pose: Tracking people by finding stylized poses",
abstract = "We develop an algorithm for finding and kinematically tracking multiple people in long sequences. Our basic assumption is that people tend to take on certain canonical poses, even when performing unusual activities like throwing a baseball or figure skating. We build a person detector that quite accurately detects and localizes limbs of people in lateral walking poses. We use the estimated limbs from a detection to build a discriminative appearance model; we assume the features that discriminate a figure in one frame will discriminate the figure in other frames. We then use the models as limb detectors in a pictorial structure framework, detecting figures in unrestricted poses in both previous and successive frames. We have run our tracker on hundreds of thousands of frames, and present and apply a methodology for evaluating tracking on such a large scale. We test our tracker on real sequences including a feature-length film, an hour of footage from a public park, and various sports sequences. We find that we can quite accurately automatically find and track multiple people interacting with each other while performing fast and unusual motions.",
author = "Deva Ramanan and Forsyth, {D. A.} and Andrew Zisserman",
year = "2005",
month = "1",
day = "1",
doi = "10.1109/CVPR.2005.335",
language = "English (US)",
isbn = "0769523722",
series = "Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005",
publisher = "IEEE Computer Society",
pages = "271--278",
booktitle = "Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005",

}

TY - GEN

T1 - Strike a pose

T2 - Tracking people by finding stylized poses

AU - Ramanan, Deva

AU - Forsyth, D. A.

AU - Zisserman, Andrew

PY - 2005/1/1

Y1 - 2005/1/1

N2 - We develop an algorithm for finding and kinematically tracking multiple people in long sequences. Our basic assumption is that people tend to take on certain canonical poses, even when performing unusual activities like throwing a baseball or figure skating. We build a person detector that quite accurately detects and localizes limbs of people in lateral walking poses. We use the estimated limbs from a detection to build a discriminative appearance model; we assume the features that discriminate a figure in one frame will discriminate the figure in other frames. We then use the models as limb detectors in a pictorial structure framework, detecting figures in unrestricted poses in both previous and successive frames. We have run our tracker on hundreds of thousands of frames, and present and apply a methodology for evaluating tracking on such a large scale. We test our tracker on real sequences including a feature-length film, an hour of footage from a public park, and various sports sequences. We find that we can quite accurately automatically find and track multiple people interacting with each other while performing fast and unusual motions.

AB - We develop an algorithm for finding and kinematically tracking multiple people in long sequences. Our basic assumption is that people tend to take on certain canonical poses, even when performing unusual activities like throwing a baseball or figure skating. We build a person detector that quite accurately detects and localizes limbs of people in lateral walking poses. We use the estimated limbs from a detection to build a discriminative appearance model; we assume the features that discriminate a figure in one frame will discriminate the figure in other frames. We then use the models as limb detectors in a pictorial structure framework, detecting figures in unrestricted poses in both previous and successive frames. We have run our tracker on hundreds of thousands of frames, and present and apply a methodology for evaluating tracking on such a large scale. We test our tracker on real sequences including a feature-length film, an hour of footage from a public park, and various sports sequences. We find that we can quite accurately automatically find and track multiple people interacting with each other while performing fast and unusual motions.

UR - http://www.scopus.com/inward/record.url?scp=24644504137&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=24644504137&partnerID=8YFLogxK

U2 - 10.1109/CVPR.2005.335

DO - 10.1109/CVPR.2005.335

M3 - Conference contribution

AN - SCOPUS:24644504137

SN - 0769523722

SN - 9780769523729

T3 - Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005

SP - 271

EP - 278

BT - Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005

PB - IEEE Computer Society

ER -