Presentation 2000/1/20
Object Recognition Based on the Analysis of Human Actions
Seiji Kimura, Michael Hild, Yoshiaki Shirai,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) It is difficult to recognize objects of similar color and texture when they are adjacent or partially occluded each other in an image. This paper describes a method for separating and recognizing objects photographed under such conditions. The method is based on the analysis of human action observed in an image sequence of the scene and the use of models of object classes expected to be present in the scene. In the first stage, an object's class and its 3-dimensional location and direction are estimated through an analysis of human action and the use of 3-D data points. In the second stage, the areas in which objects are likely to exist are delimited in key frames by fitting object models to 3-D data points. And in a third stage, exact segmentation of object regions is performed based on color and 3-D point information. As a concrete example, we study the separation and recognition of chairs and use as human action the process of a "man sits down on chair."
Keyword(in Japanese) (See Japanese page)
Keyword(in English) human action / object model / object segmentation / object recognition
Paper # PRMU99-202
Date of Issue

Conference Information
Committee PRMU
Conference Date 2000/1/20(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 Pattern Recognition and Media Understanding (PRMU)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Object Recognition Based on the Analysis of Human Actions
Sub Title (in English)
Keyword(1) human action
Keyword(2) object model
Keyword(3) object segmentation
Keyword(4) object recognition
1st Author's Name Seiji Kimura
1st Author's Affiliation Graduate School of Engineering, Osaka University()
2nd Author's Name Michael Hild
2nd Author's Affiliation Graduate School of Engineering, Osaka University
3rd Author's Name Yoshiaki Shirai
3rd Author's Affiliation Graduate School of Engineering, Osaka University
Date 2000/1/20
Paper # PRMU99-202
Volume (vol) vol.99
Number (no) 574
Page pp.pp.-
#Pages 8
Date of Issue