ACCAD - Motion Capture Lab - Research-People

Research Papers

	"Motion capture for the rest of us" Meg Geroch Journal of Computing Sciences in Colleges Volume 19 , Issue 3 (January 2004) Pages: 157 - 164 http://portal.acm.org/citation.cfm?id=948849 PDF Document "Gender Recognition from Walking Movements using Adaptive Three-Mode PCA" J. Davis and H. Gao IEEE Workshop on Articulated and Nonrigid Motion, Washington DC, June 27, 2004. PDF Document "An Expressive Three-Mode Principal Components Model for Gender Recognition" Davis, J. W., & Gao, H. Journal of Vision, 4(5), 362-377. http://www.journalofvision.org/4/5/2/ PDF Document "An Expressive Three-Mode Principal Components Model of Human Action Style" J. Davis and H. Gao, Image and Vision Computing, Vol. 21, No. 11, 2003, pp. 1001-1016. PDF Document "Recognizing Human Action Efforts: An Adaptive Three-Mode PCA Framework" J. Davis and H. Gao, International Conference on Computer Vision, Nice, France, Oct 13-16, 2003, pp. 1463-1469. PDF Document "A Reliable-Inference Framework for Pose-Based Recognition of Human Actions" J. Davis and A. Tyagi T o appear in IEEE International Conference on Advanced Video and Signal Based Surveillance, Miami, Florida, July 21-22, 2003. PDF Document "A Three-Mode Expressive Feature Model of Action Effort" J. Davis, H. Gao, and V. Kannappan IEEE Workshop on Motion and Video Computing, Orlando, Florida, December 5-6, 2002, pp. 139-144. [Received a Lockheed-Martin Best Paper Award] PDF Document "Expressive Features for Movement Exaggeration" J. Davis and V. Kannappan SIGGRAPH Conference Abstracts and Applications (Technical Sketches), San Antonio, Texas, July 24, 2002, pp. 182. PDF Document

Fellowship Paper
	The Ohio Board of Regents Incentive Fund
	"A Three-Mode Expressive Feature Model of Action Effort" by GAO, Hui Computer Vision Laboratory Dept. of Computer & Information Science The Ohio State University PDF Document Abstract Human actions can be performed with various styles. Of interest to computer vision and computer animation is a tool that can extract this stylistic information from the movement patterns to aid in analysis, synthesis, and recognition. We present a computational approach employing three-mode principal components that is capable of learning (via training data) which motion trajectories best express the differences between motion styles, so as to reliably recognize the style for new motions. We demonstrate the model by examining motion-capture walking movements that emphasize three different stylistic aspects: carrying load (light, heavy), walking pace (leisurely, in a hurry), and gender (male, female). Results show the ability of the framework to successfully model and recognize the different motion styles.

	"Human Figure Motion Synthesis and Analysis" by Arunachalam Somasundaram Advanced Computing Center for Art and Design Dept. of Computer & Information Science The Ohio State University PDF Document Abstract Realistic facial synthesis is one of the most fundamental problems in computer graphics and one of the most difficult. The complex geometric form of the human face; the underlying skeletons and muscles as well as the mechanical properties of the skin and subcutaneous layers; the coordinating brain and the uncanny ability of humans to read expressions - all make facial animation very difficult. The applications of facial modeling and animation include character animation for films, computer games, video teleconferencing, user-interface avatars, facial surgery planning etc. In this talk, I will present my ongoing research and the various issues involved in facial animation. Current motion capture systems can capture fine 3D facial movements that can be used to understand and synthesize facial motion. 3D facial movements were captured during speech and expressions using an optical motion capture system. I will describe a method to transfer motion capture marker movements to varying geometric facial mesh models. I will present a technique to add facial expressions to captured neutral speech imposing speech and geometric constraints to produce expressive speech animation and compare it with captured expressive speech. I will also present my ongoing preliminary work in developing a physically based face model and in building algorithms learnt from motion capture data to animate it. Finally, I will discuss some of the research issues involved in facial animation. Here are the links to some Movies and Pictures http://www.accad.osu.edu/~asomasun/Face/Movies http://www.accad.osu.edu/~asomasun/Face/Pictures

	"Motion capture of fluid flow (tornado, smoke, flame, etc.)" by Yootai Kim Advanced Computing Center for Art and Design Dept. of Computer Science and Engineering The Ohio State University Abstract My research goal is to control fluid simulation with video input. Animators spend a lot of time on analysing the essential features of the desired motion from video references. I want computers to do the job for fluid animaiton. I will examine the use of video to capture complex fluid motion and guide desired motion generated by an equivalent simulation. The approach will exploit vision techniques, optimization algorithms and flow simulation methods. Towards this purpose, I plan to exploit the high-resolution Vicon cameras at ACCAD to capture synchnous muliple-view video clips of fluid flow. http://windfish.cis.ohio-state.edu/tornado/tornado_report.htm