Motion Estimation via Scale-Space in Unsupervised Deep Learning.
Jaehwan KimBilel DerbelByung-Woo HongPublished in: ICOIN (2021)
Keyphrases
- scale space
- deep learning
- motion estimation
- unsupervised learning
- deep architectures
- weakly supervised
- multiscale
- scale spaces
- optical flow
- supervised learning
- machine learning
- scale selection
- image features
- video sequences
- semi supervised
- gaussian scale space
- image analysis
- image structure
- critical points
- superpixels
- super resolution
- image sequences
- keypoints
- mental models
- natural images
- dimensionality reduction
- pattern recognition
- topic models
- expectation maximization
- reinforcement learning
- text classification
- probabilistic model