Neural Marionette: Unsupervised Learning of Motion Skeleton and Latent Dynamics from Volumetric Video.
Jinseok BaeHojun JangCheol-Hui MinHyungun ChoiYoung Min KimPublished in: CoRR (2022)
Keyphrases
- unsupervised learning
- space time
- articulated human motion
- key frames
- spatial and temporal
- temporal filtering
- object motion
- input video
- motion features
- video frames
- video sequences
- dynamic scenes
- static images
- human motion
- motion model
- motion capture
- temporal consistency
- moving camera
- temporal continuity
- low frame rate
- multimedia
- image sequences
- successive frames
- optical flow
- camera motion
- dynamic textures
- supervised learning
- visual cues
- video content
- motion analysis
- network architecture
- motion patterns
- machine learning
- video shots
- visual data
- motion field
- video data
- neural network
- layered representation
- object recognition
- video objects
- motion estimation
- video surveillance
- action recognition
- dimensionality reduction
- motion capture data
- wide area motion imagery
- recurrent networks
- moving objects
- motion trajectories
- video streams
- video analysis
- semi supervised
- expectation maximization
- reinforcement learning
- three dimensional
- human actions