Action Segmentation for RGB Video Frames Using Skeleton 3D Data of NTURGB+D

Abstract

Action segmentation or video segmentation which used to extract action from video frames. It plays role in various applications, i.e., visual effect assistance in the movies, scene understanding in detail, virtual background creation, and a design CAD system that can identify automatically human action from videos without any object interference. This paper presents a system that automatically segments actions from videos. The window size is variable and depends on input video. The dataset used to show experimental data is NTURGB+D. The action segmentation has been shown using 3D skeleton information on RGB videos of NTURGB+D. The experimental results have shown the performance and it test results on 5 random action videos.