Backing Off: Hierarchical Decomposition of Activity for 3D Novel Pose Recovery

[edit]

John Darby
Baihua Li
Nicholas Costen
David J. Fleet
Neil D. Lawrence, University of Sheffield

in British Machine Vision Conference

Related Material

Abstract

For model-based 3D human pose estimation, even simple models of the human body lead to high-dimensional state spaces. Where the class of activity is known a priori, lowdimensional activity models learned from training data make possible a thorough and efficient search for the best pose. Conversely, searching for solutions in the full state space places no restriction on the class of motion to be recovered, but is both difficult and expensive. This paper explores a potential middle ground between these approaches, using the hierarchical Gaussian process latent variable model to learn activity at different hierarchical scales within the human skeleton. We show that by training on full-body activity data then descending through the hierarchy in stages and exploring subtrees independently of one another, novel poses may be recovered. Experimental results on motion capture data and monocular video sequences demonstrate the utility of the approach, and comparisons are drawn with existing low-dimensional activity models


@InProceedings{darby-backing09,
  title = 	 {Backing Off: Hierarchical Decomposition of Activity for 3D Novel Pose Recovery},
  author = 	 {John Darby and Baihua Li and Nicholas Costen and David J. Fleet and Neil D. Lawrence},
  booktitle = 	 {British Machine Vision Conference},
  year = 	 {2009},
  month = 	 {00},
  edit = 	 {https://github.com/lawrennd//publications/edit/gh-pages/_posts/2009-01-01-darby-backing09.md},
  url =  	 {http://inverseprobability.com/publications/darby-backing09.html},
  abstract = 	 {For model-based 3D human pose estimation, even simple models of the human body lead to high-dimensional state spaces. Where the class of activity is known a priori, lowdimensional activity models learned from training data make possible a thorough and efficient search for the best pose. Conversely, searching for solutions in the full state space places no restriction on the class of motion to be recovered, but is both difficult and expensive. This paper explores a potential middle ground between these approaches, using the hierarchical Gaussian process latent variable model to learn activity at different hierarchical scales within the human skeleton. We show that by training on full-body activity data then descending through the hierarchy in stages and exploring subtrees independently of one another, novel poses may be recovered. Experimental results on motion capture data and monocular video sequences demonstrate the utility of the approach, and comparisons are drawn with existing low-dimensional activity models},
  key = 	 {Darby:backing09},
  linkpdf = 	 {ftp://ftp.dcs.shef.ac.uk/home/neil/BMVC2009CR06.pdf},
  linksoftware = {http://www.docm.mmu.ac.uk/STAFF/J.Darby/code.htm},
  OPTgroup = 	 {}
 

}
%T Backing Off: Hierarchical Decomposition of Activity for 3D Novel Pose Recovery
%A John Darby and Baihua Li and Nicholas Costen and David J. Fleet and Neil D. Lawrence
%B 
%C British Machine Vision Conference
%D 
%F darby-backing09	
%P --
%R 
%U http://inverseprobability.com/publications/darby-backing09.html
%X For model-based 3D human pose estimation, even simple models of the human body lead to high-dimensional state spaces. Where the class of activity is known a priori, lowdimensional activity models learned from training data make possible a thorough and efficient search for the best pose. Conversely, searching for solutions in the full state space places no restriction on the class of motion to be recovered, but is both difficult and expensive. This paper explores a potential middle ground between these approaches, using the hierarchical Gaussian process latent variable model to learn activity at different hierarchical scales within the human skeleton. We show that by training on full-body activity data then descending through the hierarchy in stages and exploring subtrees independently of one another, novel poses may be recovered. Experimental results on motion capture data and monocular video sequences demonstrate the utility of the approach, and comparisons are drawn with existing low-dimensional activity models
TY  - CPAPER
TI  - Backing Off: Hierarchical Decomposition of Activity for 3D Novel Pose Recovery
AU  - John Darby
AU  - Baihua Li
AU  - Nicholas Costen
AU  - David J. Fleet
AU  - Neil D. Lawrence
BT  - British Machine Vision Conference
PY  - 2009/01/01
DA  - 2009/01/01	
ID  - darby-backing09	
SP  - 
EP  - 
L1  - ftp://ftp.dcs.shef.ac.uk/home/neil/BMVC2009CR06.pdf
UR  - http://inverseprobability.com/publications/darby-backing09.html
AB  - For model-based 3D human pose estimation, even simple models of the human body lead to high-dimensional state spaces. Where the class of activity is known a priori, lowdimensional activity models learned from training data make possible a thorough and efficient search for the best pose. Conversely, searching for solutions in the full state space places no restriction on the class of motion to be recovered, but is both difficult and expensive. This paper explores a potential middle ground between these approaches, using the hierarchical Gaussian process latent variable model to learn activity at different hierarchical scales within the human skeleton. We show that by training on full-body activity data then descending through the hierarchy in stages and exploring subtrees independently of one another, novel poses may be recovered. Experimental results on motion capture data and monocular video sequences demonstrate the utility of the approach, and comparisons are drawn with existing low-dimensional activity models
ER  -

Darby, J., Li, B., Costen, N., Fleet, D.J. & Lawrence, N.D.. (2009). Backing Off: Hierarchical Decomposition of Activity for 3D Novel Pose Recovery. British Machine Vision Conference :-