Open-world driving scene segmentation via multi-stage and multi-modality fusion of vision-language embedding.
Yingjie NiuMing DingYuxiao ZhangMaoning GeHanting YangKazuya TakedaPublished in: IV (2023)
Keyphrases
- multistage
- multi modality
- scene segmentation
- closed world
- open world
- multi modal
- medical images
- information theoretic
- image registration
- mutual information
- data exchange
- incomplete information
- dynamic programming
- indoor scenes
- object tracking
- vision system
- imaging modalities
- computer vision
- single image
- image processing
- stable model semantics
- machine learning
- optimal policy
- deformation field
- high level
- dynamic environments
- query answering
- knowledge base