DiPE: Deeper into Photometric Errors for Unsupervised Learning of Depth and Ego-motion from Monocular Videos.
Hualie JiangLaiyan DingRui HuangPublished in: CoRR (2020)
Keyphrases
- ego motion
- unsupervised learning
- moving camera
- dynamic scenes
- field of view
- d scene
- camera motion
- static scene
- image sequences
- optical flow
- stereo camera
- video sequences
- visual odometry
- depth information
- image velocity
- robust estimation
- depth map
- motion parameters
- flow field
- optic flow
- multi camera
- semi supervised
- multi view
- video data
- photometric stereo
- expectation maximization
- video surveillance
- motion field
- object recognition
- velocity field
- feature selection
- least squares
- space time
- video camera
- probabilistic model
- single camera
- background subtraction
- high quality
- computer vision
- machine learning