“Content-Preserving Motion Stylization using Variational Autoencoder” by Liao, Kim, Koike and Hwang
Conference:
Type(s):
Entry Number: 03
Title:
- Content-Preserving Motion Stylization using Variational Autoencoder
Session/Category Title: Posters: Animation & Simulation
Presenter(s)/Author(s):
Abstract:
This work proposes a novel motion style transfer network that transfers motion styles among different motion categories using the concept of the variational autoencoder. The proposed network is effective in transferring style among various motion categories and is also able to create stylized motion that is unseen in the dataset.
