Human Motion Imagination and Prediction- A Survey

Wafaa Shihab Ahmed 1*, Abdul amir A. Karim 2

1Department of Computer Science, University of Technology, Baghdad, Iraq
2 Department of Computer Science, University of Technology, Baghdad, Iraq
*Corresponding Author: 111798@student.uotechnology.edu.iq

DOI: 10.18081/2226-3284/5-10/30-45

AbstractKey words
Human motion generation and prediction is one of important subjects in computer vision, human robot interactions and animations. There are many methods have been used for human motion modelling. This paper illustrate a survey about human motion imagination and prediction and explain the types and methods of video generation and human motion modelling. In this paper the Recurrent Neural Network/ Long Short Term Memory (RNN/LSTM), Generative Adversarial Network (GAN) and Variational Auto Encoders (VAE) models have been introduced. These models have been used for generating the spatial-temporal cuboids or for predicting the intensity pixels trajectory in the scene and give good results with short term prediction. This paper also shows the perceptual quality metrics that used for computing the performance of the methods and techniques which has been used for modelling.
Video generation, Human motion imagination, Human motion prediction, Perceptual quality metrics.

Download full article/PDF

Leave a Reply

Your email address will not be published. Required fields are marked *