G3AN: Disentangling Appearance and Motion for Video Generation

Abstract

Creating realistic human videos entails the challenge of being able to simultaneously generate both appearance, as well as motion. To tackle this challenge, we introduce G${}^3$AN, a novel spatio-temporal generative model, which seeks to capture the distribution of high dimensional video data and to model appearance and motion in disentangled manner. The latter is achieved by decomposing appearance and motion in a three-stream Generator, where the main stream aims to model spatio-temporal consistency, whereas the two auxiliary streams augment the main stream with multi-scale appearance and motion features, respectively. An extensive quantitative and qualitative analysis shows that our model systematically and significantly outperforms state-of-the-art methods on the facial expression datasets MUG and UvA-NEMO, as well as the Weizmann and UCF101 datasets on human action. Additional analysis on the learned latent representations confirms the successful decomposition of appearance and motion. Source code and pre-trained models are publicly available.

arXiv commentCVPR 2020, project link https://wyhsirius.github.io/G3AN/

References in corpus (18)

• Adam: A Method for Stochastic Optimization
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
• GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
• Large Scale GAN Training for High Fidelity Natural Image Synthesis
• Spectral Normalization for Generative Adversarial Networks
• Self-Attention Generative Adversarial Networks
• Progressive Growing of GANs for Improved Quality, Stability, and Variation
• Improved Techniques for Training GANs
• Generating Videos with Scene Dynamics
• GAN Dissection: Visualizing and Understanding Generative Adversarial Networks
• Temporal Generative Adversarial Nets with Singular Value Clipping
• JointGAN: Multi-Domain Joint Distribution Learning with Generative Adversarial Nets
• Disentangled Person Image Generation
• Guided Image-to-Image Translation with Bi-Directional Feature Transformation
• Video Generation from Single Semantic Label Map
• Pose Guided Human Video Generation
• Flow-Grounded Spatial-Temporal Video Prediction from Still Images
• Image Generation from Layout