HUMANISE: Language-conditioned Human Motion Generation in 3D Scenes.
Zan WangYixin ChenTengyu LiuYixin ZhuWei LiangSiyuan HuangPublished in: NeurIPS (2022)
Keyphrases
- human motion
- d scene
- motion capture
- image sequences
- human body
- video sequences
- depth map
- optical motion capture
- single image
- spatio temporal
- body parts
- motion synthesis
- moving camera
- optical flow
- scene structure
- image based rendering
- motion recognition
- motion capture data
- view dependent
- motion sequences
- human actions
- photorealistic
- human pose
- three dimensional
- articulated human motion
- machine learning
- state space
- computer vision