Domain Adaptive Semantic Segmentation with Self-Supervised Depth Estimation.
Qin WangDengxin DaiLukas HoyerLuc Van GoolOlga FinkPublished in: ICCV (2021)
Keyphrases
- street scenes
- depth estimation
- semantic segmentation
- stereo vision
- depth map
- dynamic scenes
- stereo matching
- conditional random fields
- superpixels
- real scenes
- depth information
- scene classification
- stereo pair
- multi view
- scene understanding
- disparity map
- weakly supervised
- object segmentation
- object classes
- feature matching
- three dimensional
- specular reflection
- object categories
- multi class
- high quality
- training data