This preliminary study presents a system capable of
recognizing human gesture in real-time. The gesture is acquired from a Kinect2 sensor which provides skeleton joints represented by three-dimensional coordinate points. The model set consists of eight motion gestures is provided for basis of gesture recognition using Dynamic Time Warping (DTW) algorithm. DTW algorithm is utilized to identify in real time manner by measuring the shortest combined distances in x, y, and z coordinates in order to determined the matched gesture. It can be shown that the system
is able to recognize these 8 motions in real time with some
limitations. The findings of the this study will provide solid foundation of further research in which the ultimate goal of the research is to create system to automatically recognize sequence of motions in Indonesian traditional dances and convert them into standardized Resource Description Framework (RDF) scripts for the purpose of preserving these dances.