Joint Unsupervised Learning of Optical Flow and Depth by Watching Stereo Videos.
Yang WangZhenheng YangPeng WangYi YangChenxu LuoWei XuPublished in: CoRR (2018)
Keyphrases
- optical flow
- unsupervised learning
- stereo disparity
- stereoscopic video
- real time stereo
- depth map
- pre recorded
- moving camera
- depth information
- motion features
- depth estimation
- image sequences
- depth cues
- motion cues
- image motion
- d scene
- occlusion boundaries
- disparity estimation
- supervised learning
- rigid body motion
- computer vision
- flow field
- scene flow
- binocular disparity
- ego motion
- optical flow estimation
- free viewpoint
- optical flow computation
- depth images
- dynamic scenes
- video sequences
- depth data
- motion segmentation
- view synthesis
- multi view
- video frames
- stereo matching
- stereo pair
- motion field
- machine learning
- three dimensional
- semi supervised
- stereo vision
- motion parallax
- stereo images
- motion estimation
- feature selection
- optical flow algorithm
- camera motion
- time of flight
- motion parameters
- early vision
- dimensionality reduction
- motion model
- stereo camera
- model selection
- moving objects
- multi camera
- object recognition
- learning algorithm
- video content
- image pairs
- video clips
- video data
- feature points
- super resolution