SME-Net: Sparse Motion Estimation for Parametric Video Prediction Through Reinforcement Learning.
Yung-Han HoChuan-Yuan ChoGuo-Lun JinWen-Hsiao PengPublished in: ICCV (2019)
Keyphrases
- motion estimation
- reinforcement learning
- video sequences
- video data
- video compression standard
- prediction accuracy
- video frames
- temporal filtering
- prediction error
- video content
- video compression
- video streams
- motion vectors
- space time
- prediction model
- inter frame
- real time
- motion compensated
- function approximation
- video coding
- video encoder
- block matching
- video processing
- multimedia
- real time face tracking
- video analysis
- state space
- image sequences
- learning algorithm
- temporal redundancy
- phase correlation
- reinforcement learning algorithms
- neural network
- motion segmentation
- motion field
- markov decision processes
- sparse representation
- computer vision
- video retrieval
- video shots
- digital video
- model free
- video surveillance
- compression efficiency
- motion compensation
- real time video
- event detection
- machine learning