A Study on Basic Human Activity Recognition Based on Sensor Jacket for Interaction in Metaverse Environment

Abstract

Metaverse is attracting attention as activities in various fields online have increased recently. On the metaverse platform, users express their feelings and communicate with other users using gestures, which are nonverbal communication, as well as voice and text messages. In this study, we constructed a basic dataset by utilizing the behavior used in the metaverse platform. Previously, motion capture systems were used to obtain accurate motion data. However, there is a problem that accessibility and convenience are poor due to the complexity of expensive equipment and use. To compensate for this problem, we have created a sensor jacket for wearing and fast motion recognition. Therefore, to verify recognition accuracy for sensor jacket use in metaverse environments, motion data from motion capture systems and sensor jacket systems are utilized by extreme learning machine algorithms and motion-sphere to compare and analyze recognition and matching rates. In addition, an easy-to-use motion recognition sensor jacket is proposed to communicate smoothly in a metaverse environment.