Posture corpus for the behavior generator; an expressive virtual audience with flexible behavioral styles

To create a virtual audience with parameterized behavioral styles, real audiences’s behavior was observed, and data of their personality, mood, attitude, and energy level was collected in four conditions: positive presentation, neutral presentation, boring presentation, and critical presentation. The audience behavior was then coded for the preparation of modeling the behavioral patterns. This corpus has been used in the paper: N. Kang, W.-P. Brinkman, B. Van Riemsdijk, and M. A. Neerincx, “An Expressive Virtual Audience with Flexible Behavioral Styles,” IEEE Trans. Affect. Comput., vol. 4, no. 4, pp.