Simoun: Synergizing Interactive Motion-appearance Understanding for Vision-based Reinforcement Learning.
Yangru HuangPeixi PengYifan ZhaoYunpeng ZhaiHaoran XuYonghong TianPublished in: ICCV (2023)
Keyphrases
- reinforcement learning
- static images
- object motion
- motion cues
- motion analysis
- motion estimation
- image sequences
- motion model
- space time
- human computer interaction
- human motion
- appearance model
- appearance features
- vision sensors
- moving objects
- dynamic textures
- motion parameters
- user friendly
- real time
- augmented reality
- virtual reality
- optical flow
- vision system
- feature points
- camera motion
- layered representation
- motion tracking
- motion planning
- motion patterns
- motion segmentation
- computer graphics
- user interaction
- observation model
- state space
- function approximation
- rotational motion
- high degree of freedom