Real-time analysis of expressive cues in human movement

This paper presents results from our research concerning the development and experimenting of computational models for real-time analysis of expressive intentions in human movement. We aim at discovering and measure movement cues useful for recognizing expressive content in movement, and in particular from dancers, music performers, actors on a theatre stage, or visitors in a museum exhibit. The ultimate goal is to contribute toward a deeper understanding of the relations between gesture, music, and visual languages, and enable artists with conceptual as well as system tools for interactive performance. In a broader perspective, we aim at enhancing man-machine communication by adding a new channel: expressiveness. The analysis method and computational models are inspired to several sources, including humanistic theories on non-verbal communication, such as Rudolf Laban's Theory of Effort. In order to test the algorithms developed in the project, we created, with the help of professional choreographers, a library of dance fragments, called microdances each one performed with different expressive content. Furthermore, we analyze and compare different dance fragments performed with the same expressive content, in order to understand invariants and cues enabling the recognition of such commonalities. Evaluation of the results obtained with spectators’ ratings is also part of the project. Research outputs are implemented in our EyesWeb open software platform. This research is partially funded by the EU IST-20410 Project MEGA (Multisensory Expressive Gesture Applications).