Bootstrapped Self-Supervised Training with Monocular Video for Semantic Segmentation and Depth Estimation.
Yihao ZhangJohn J. LeonardPublished in: IROS (2021)
Keyphrases
- street scenes
- depth estimation
- semantic segmentation
- dynamic scenes
- stereo vision
- depth map
- video sequences
- scene understanding
- image sequences
- stereo matching
- depth information
- superpixels
- training set
- conditional random fields
- video surveillance
- video frames
- stereo pair
- feature matching
- super resolution
- training examples
- multi view
- real scenes
- disparity map
- graphical models
- pose estimation
- specular reflection
- object segmentation
- pairwise
- input image
- motion segmentation
- three dimensional
- semi supervised