Action Club Seminar: Machine Learning for Human Motion Analysis and Synthesis

September 9, 2016 @ 04:45 pm to 06:00 pm

Yanxi Liu, Penn State

127 Noll Conference Room

From human gait to dance, human movements provide rich, complex yet coherent spatiotemporal data that can be characteristic of individual subjects. We develop computer algorithms to automatically learn such discriminative features from digitized data (video and motion capture data). In this talk, I present two projects: (1) Gait analysis from video data: we use a novel gait sequence representation based on mathematical frieze patterns. (2) Dance analysis and synthesis via on-line human perception using motion capture data: we explore the complex relationship between perceived dance quality/dancer's gender and dance movements respectively. I shall also discuss the feasibility of utilizing these technologies in our current project of Taiji + Advanced Technology for Smart Health.

Contact

Bob Sainburg
rls45@psu.edu