Abstract
This paper shows a method to identify a learner's action simply by a few data from a stereo vision camera system. It is assumed that the learner's head and both hands move characteristically while (s)he is learning in a seated position. In this method, first, the motions of the three body parts of the learner are measured. Then the behavior is identified by comparing the frequency characteristic of the motion data with that of pre-defined behaviors. We examined this method on nine typical learning actions such as “horizontal writing”, “document creation with word-processing software” and so on. As a result, it has been proved that the method is effective to identify learner's
action on instrumental grounds.
