Unsupervised Optical Flow Using Cost Function Unrolling.
Gal LifshitzDan RavivPublished in: CoRR (2020)
Keyphrases
- cost function
- optical flow
- image sequences
- least squares
- objective function
- optical flow estimation
- unsupervised learning
- supervised learning
- dense optical flow
- vector field
- d scene
- motion model
- expected cost
- semi supervised
- motion estimation
- moving objects
- optical flow computation
- motion field
- flow field
- motion parameters
- image motion
- motion discontinuities
- machine learning
- supervised classification
- moving camera
- structure from motion
- ego motion
- global minimum
- data driven
- active learning
- feature tracking
- unsupervised manner
- feature selection
- scene flow
- computer vision
- data sets
- completely unsupervised
- binocular disparity