SYSTEM FOR RECOGNIZING GESTURE BASED ON DEEP LEARNING USING SENSOR DATA AND METHOD THEREOF

摘要

The present invention relates to a system for recognizing a gesture based on deep learning using sensor data and a method thereof. According to the present invention, the method comprises the steps of: sequentially receiving sensor data of X, Y, and Z axes of an acceleration sensor and X, Y, and Z axes of an angular velocity sensor and extracting a data section through a sliding window; forming a multi-layer by a max-pooling method to learn a robust spatial feature; learning the spatial feature through a convolutional operation and connecting a convolutional output of the multilayer in order to extract the spatial feature of the multilayer; learning relationship and a temporal feature of sequential data using features of each of the axes of the sensor data; and learning the spatial and temporal features through full connection. According to the present invention, an operation pattern of a person can be recognized by using sensor data information on a wireless device such as a smart band. In addition, the operation pattern can be more precisely recognized by using spatial and temporal features.;COPYRIGHT KIPO 2019