src="/
  • Ultimate Supporter Ultimate Supporter src="/
  • Ultimate Attendee Ultimate Attendee

Date: Sunday, December 13th
Time: 9:00am - 10:00am
Venue: Main Room
Session Chair:


Note: All live sessions will be screened on Singapore Time/GMT+8. Convert your time zone here.


Q&A for ZooBuilder: 3D Animation Generation from Videos for Quadrupeds

Author(s)/Presenter(s):
Abassin Sourou Fangbemi, Ubisoft China, China
Alexis Rolland, Ubisoft China, China

Abstract: With the objective of automating the creation of animations for wildlife, this project from Ubisoft China introduces a novel strategy for generating synthetic training data for 2D and 3D pose estimation of animals using keyframe animations. The machine learning models trained with this synthetic data are integrated into an end-to-end pipeline called ZooBuilder, that takes a video input of an animal in the wild and produces the corresponding 3D animation. With this approach, we produce motion capture-like data from videos that can be used to accelerate the workflow of animators.


Back