Recognition of Hand Raising Gestures for a Remote Learning Application

A central technical issue in developing synchronous distance learning technology is enabling the remote class and the instructor to interact with each other. Issues such as "how does a student capture the instructor\´s attention?", "how can the instructor select one student to converse with?", and "how can the instructor attend to the student once (s)he has been selected?" are complex problems that must be addressed if the class and instructor are to interact in an effective manner. This paper describes the use of Hidden Markov Models for the recognition of students signaling their intent to interact with the instructor using "traditional" classroom hand gestures such as raising and waving hand motions. Hand raising gestures are detected using motion cues over a sequence of omni-directional images using a set of pre-defined Hidden Markov Models.