A Method of Recognizing Body Movements Based on a Self-viewpoint Video

The most critical human sensory function resides in vision.This paper focuses on utilizing visual information, specifically self-perspective footage, to identify individual movements.Existing researches require third-party filming to recognize human body movements and states.The proposed method, on the other hand, simply attaches a camera to the human head and enables the recognition of the subject's actions.Consequently, it becomes easier to monitor daily movements of a human and gather his/her data on body kinetics.This approach would be beneficial in scenarios involving individuals engaging in risky behavior or, during a certain emergency, providing valuable assistance.