VideoMocap: Modeling Physically Realistic Human Motion from Monocular Video Sequences

by sgallege on January 23, 2011

in Assignment 1

Assigned Paper:

VideoMocap: Modeling Physically Realistic Human Motion from Monocular Video Sequences

1. Sentence:

In this paper, we present a video-based motion capture technique for modeling physically realistic 3D human motion from uncalibrated monocular video sequences such as sports video footage

2. Problem:

Current motion capture technologies are often restrictive, cumbersome, and expensive. Optical and magnetic motion capture systems must be operated in carefully calibrated, restrictive lab settings, inhibiting the possibility of acquiring outdoor activities. Inertial or mechanical systems, on the other hand, are not constrained by a fixed capture space, but require the subject to wear cumbersome sensors or confined exo -skeletons,reducing the naturalness and quality in the performance.

3. Key Idea:

Combine the power of automatic computer vision techniques and physics-based motion modeling techniques to generate 3D human motion with a high degree of physical realism.

4.a What the paper does:

Provides a system that models physically realistic human motion from uncalibrated monocular video sequences. Utilizing following techniques: key-frame based multi-joint tracking, Motion Interpolation, Motion refinement.

4.b. What it could be used for:

Motion capture animation from single camera footage. Create new viewpoints from the model.

5. Resources:

Video http://www.youtube.com/watch?v=EjpS5WD6V94

Previous post:

Next post: