3D GESTURE RECOGNITION SYSTEM AND METHOD THEREOF AND COMPUTER PROGRAM THEREFOR

摘要

A three-dimensional (3D) gesture recognition system includes: a client device configured to receive an input vector including location information corresponding to a users 3D gesture and process the input vector to generate data obtained by normalization of dimensions and numbers of the input vector; and an asynchronous server configured to receive the data from the client device, generate a result value of recognizing the users gesture from the data by applying a machine learning model, and transmit the result value to the client device. According to the 3D gesture recognition system, the input of a form of a continuous and variable 3D vector is normalized through preprocessing, a machine learning algorithm is processed through the asynchronous server, and a machine learning model appropriate for environments and resources is constructed. Accordingly, it is possible to obtain a gesture recognition result at a high speed without deteriorating performance such as frame drops.;COPYRIGHT KIPO 2020