MOSS: Motion-based 3D Clothed Human Synthesis from Monocular Video.
Hongsheng WangXiang CaiXi SunJinhong YueShengyu ZhangFeng LinFei WuPublished in: CoRR (2024)
Keyphrases
- image sequences
- motion capture data
- static images
- space time
- video sequences
- articulated human motion
- moving camera
- object motion
- human motion
- temporal filtering
- motion capture
- visual cues
- optical flow
- camera motion
- ego motion
- spatial and temporal
- depth perception
- input video
- human movements
- independently moving objects
- motion estimation
- visual data
- human activities
- motion features
- multiple visual cues
- motion patterns
- video content
- dynamic scenes
- motion analysis
- video frames
- motion trajectories
- human actions
- motion model
- motion history images
- multimedia
- moving objects
- dynamic textures
- temporal coherence
- temporal continuity
- feature points
- action recognition
- key frames
- image motion
- pose estimation
- layered representation
- relative depth
- computer vision
- perspective images
- video data
- projective geometry
- motion field
- video scene
- human pose
- motion parameters
- motion segmentation
- video shots
- autonomous navigation
- mechanical devices
- human body
- spatio temporal