Project

Rapidnition: Rapid User-Customizable Gesture Recognition

Groups

Rapidnition is a new way of thinking about gesturally controlled interfaces. Rather than forcing users to adapt their behavior to a predefined gestural interface, Rapidnition frees users to define their own gestures, which the system rapidly learns. The machine learning algorithms at the core of Rapidnition enable it to quickly infer a user�s gestural vocabulary, using a small number of user-demonstrated examples of each gesture. Rapidnition is capable of recognizing not just static postures but also dynamic temporal gestures. In addition, Rapidnition allows the user to define complex, nonlinear, continuous-mapping spaces. Rapidnition is currently being applied to the real-time recognition of musical gestures to rigorously test both the discrete and continuous recognition abilities of the system.