Prediction of Human Body Motion from Video Sequences

Learning to predict human body motion has emerged as a meaningful research in computer vision and artificial intelligence. This paper presents the study on predicting human body motion from video sequences. We propose a human body motion prediction network integrating the recent advanced 2D feature extraction and video sequences prediction. Based on the temporal characteristics extracted from video sequences, our network realizes the prediction of the human motion. We train the network using the video based human pose datasets and demonstrate good performance of our network on 2D human body motion prediction through quantitative and qualitative results. Experimental results prove the feasibility of our method.