Online feedback system for public speakers

摘要

The development of Affective Computing has witnessed tremendous number of studies about facial and vocal expression, while bodily expression only comprises the minority. However, with the emerging of social signal processing, people have paid more attention to the significant role of body language in social communication and emotional expression. One situation when body language shows its importance is public speaking. This paper presents an online feedback system for public speakers, in which emotion recognised from body language of speakers is regarded as the primary component for analysis. The system captures input through a Microsoft Kinect, thus offer users with a convenience way to interact. In order to recognise bodily expression, a posture and gesture representation method based on Laban Movement Analysis was adopted. The system could be used as a support tool for people who learn presentation skills, as well as for public speakers who need immediate feedbacks. This paper will present the architecture of the system, along with details about the employed methods and a collected database of presentations.