深層学習を用いたグループワーク中の重要動作の検出

抄録

This study aims to support appropriate facilitation by informing teachers of students ’learning situations during group work, a form of active learning. We investigated the possibility of detecting important body movements associated with learning using sensors embedded in smartphones worn by learners. As a result, although the accuracy of the multi-class classification of multiple important actions was not sufficient, the accuracy of the classification using individual detectors was improved mainly for nodding. Combining these individual detectors may lead to the construction of a support system that detects multiple important actions and conveys the information to the facilitator.