Radio Calisthenics Motion Detection and Diagnosis Using Recurrent Neural Network
Masato Sakai, Youji Ochi · 2018
Research that supports physical actions requires motion detection and identification. To analyze and analyze the behavior quantitatively, specifically, motion identification requires the definition of a feature quantity from input data and examination of a model of identification. However, this may be difficult in some behavior patterns. Therefore, in this research, we use RNN (Recurrent Neural Network) to construct physical identifiers obtained from Kinect as time series data to develop the motion detection. We apply the detection method to the radio calisthenics movement in an evaluation experiment to verify the usefulness of the method.