Exploiting Motion Capture for Virtual Human Animation Data Collection and Annotation Visualization

Motion capture (mocap) provides highly precise data of human movement which can be used for empirical analysis and virtual human animation. In this paper, we describe a corpus that has been collected for the purpose of modelling movement in a dyadic conversational context. We describe the technical setup, scenarios and challenges involved in capturing the corpus, and present ways of annotating and visualizing the data. For visualization we suggest the techniques of motion trails and animated re-creation. We have incorporated these motion capture visualization techniques as extensions to the ANVIL tool and into a procedural animation system, and show a first attempt at automated analysis of the data (handedness detection).