Inference of Human Postures by Classification of 3D Human Body Shape

  • Authors:
  • Isaac Cohen;Hongxia Li

  • Affiliations:
  • -;-

  • Venue:
  • AMFG '03 Proceedings of the IEEE International Workshop on Analysis and Modeling of Faces and Gestures
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper we describe an approach for inferring thebody posture using a 3D visual-hull constructed from aset of silhouettes. We introduce an appearance-based,view-independent, 3D shape description for classifyingand identifying human posture using a support vectormachine. The proposed global shape description isinvariant to rotation, scale and translation and variescontinuously with 3D shape variations. This shaperepresentation is used for training a support vectormachine allowing the characterization of human bodypostures from the computed visual hull. The mainadvantage of the shape description is its ability to capturehuman shape variation allowing the identification of bodypostures across multiple people. The proposed method isillustrated on a set of video streams of body posturescaptured by four synchronous cameras.