Presentation 2004-07-16
Static Image Based Human Pose Recovery by Optimal Candidate Decision
Hui CAO, Toshimitsu TANAKA, Noboru SUGIE,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We proposed a new approach to recovering human pose from a static image via optimal candidate pose decision. It includes two steps: generating candidate poses and deciding the most probable one. For a given image input, a set of candidate poses are generated by a combination algorithm-selecting seeds out of training data set that are close to image input, and following with exhaustive combinations of pose parts of selected seeds. The most probable pose is then selected from the candidate poses depending on a joint-data subspace based decision model, where the joint-data subspace characterizes the correlation of image and pose. The embedded combination ability can not only augment the candidate poses but also avoid generating nonexistent poses that are usually occurred when using inference method. The experiments results including synthetic and real images show good performances achieved by the proposed method.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Human pose estimation / bayesian decision / parts combination / weighted subspace
Paper # IE2004-28,PRMU2004-52,MVE2004-29
Date of Issue

Conference Information
Committee MVE
Conference Date 2004/7/9(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Media Experience and Virtual Environment (MVE)
Language ENG
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Static Image Based Human Pose Recovery by Optimal Candidate Decision
Sub Title (in English)
Keyword(1) Human pose estimation
Keyword(2) bayesian decision
Keyword(3) parts combination
Keyword(4) weighted subspace
1st Author's Name Hui CAO
1st Author's Affiliation Graduate School of Information Science, Nagoya University()
2nd Author's Name Toshimitsu TANAKA
2nd Author's Affiliation Graduate School of Science and Technology, Meijo University
3rd Author's Name Noboru SUGIE
3rd Author's Affiliation Graduate School of Science and Technology, Meijo University
Date 2004-07-16
Paper # IE2004-28,PRMU2004-52,MVE2004-29
Volume (vol) vol.104
Number (no) 196
Page pp.pp.-
#Pages 5
Date of Issue