Unsupervised Learning of Depth and Ego-Motion From Monocular Video Using 3D Geometric Constraints.
Reza MahjourianMartin WickeAnelia AngelovaPublished in: CVPR (2018)
Keyphrases
- ego motion
- geometric constraints
- camera motion
- epipolar geometry
- moving camera
- video sequences
- dynamic scenes
- field of view
- image sequences
- d scene
- optical flow
- stereo camera
- robust estimation
- scene structure
- camera calibration
- image motion
- depth information
- video data
- motion parameters
- multi camera
- optic flow
- multiple images
- flow field
- structure from motion
- feature points
- velocity field
- bundle adjustment
- video surveillance
- motion field
- background subtraction
- moving objects
- space time
- multi view
- multiple cameras
- computer vision
- motion vectors
- stereo pair
- vision system
- active learning
- image processing