GAM-Depth: Self-Supervised Indoor Depth Estimation Leveraging a Gradient-Aware Mask and Semantic Constraints.
Anqi ChengZhiyuan YangHaiyue ZhuKezhi MaoPublished in: ICRA (2024)
Keyphrases
- depth estimation
- semantic constraints
- depth map
- stereo vision
- depth information
- semantic knowledge
- stereo matching
- semantic representation
- dynamic scenes
- scene understanding
- dense depth maps
- real scenes
- knowledge sources
- noun phrases
- stereo pair
- super resolution
- feature matching
- multi view
- d scene
- depth from defocus
- high quality
- stereo images
- relational schema
- low resolution
- data sets
- image sequences
- vision system
- training data
- three dimensional
- fuzzy relations
- high resolution
- input data
- disparity map
- post processing
- surface reconstruction
- semantic information
- augmented reality