I s a a c C O H E N
3D Human body modeling
D e s c r i p t i o n
Visual sensing of humans allows the computer to derive a contextual
awareness that is useful for understanding the person's
activity. Interpreting the person's gaze, facial expression, limbs
motions and hand gestures provide a variety of information cues for
characterizing the person's physical and emotional
state. Understanding human action in an environment is a challenging
task, as it involves different granularities in its analysis and
description, according to the targeted application.
In this research effort we address the problem of accurate 3D body
shape and motion description from a set of video sensors for the
understanding and recognition of human body postures, its kinematics
and the corresponding gestures.
S t u d e n t s
- Mun Wai Lee
R e s u l t s
Tracking Articulated Body Model from Multiple Synchronized Videos
|
| Projected 3D model onto one of the 3 views used |
P u b l i c a t i o n s
- Human Body Tracking with Auxiliary Measurements IEEE International Workshop on Analysis and Modeling of Faces and Gestures, ICCV 2003
- Particle Filter with analytical inference for human body tracking, IEEE Workshop on Motion and Video Computing, Orlando, Florida. 5-6 December, 2002.
- 3D Body Reconstruction for Immersive Interaction, Second InternationalWorkshop on Articulated Motion and Deformable Objects Palma de Mallorca, Spain, 21-23 November, 2002.
- Human Body Posture Inference for Immersive Interaction, ACM, International Workshop on Immersive Telepresence , ACM Multimedia 2002, Juan Les Pins, France. 6 December, 2002.