SfM-TTR: Using Structure from Motion for Test-Time Refinement of Single-View Depth Networks.
Sergio IzquierdoJavier CiveraPublished in: CVPR (2023)
Keyphrases
- structure from motion
- single view
- multi view
- bundle adjustment
- camera pose
- depth map
- image sequences
- perspective projection
- multiple views
- camera parameters
- point cloud
- motion stereo
- computer vision
- missing data
- scene depth
- feature tracking
- motion estimates
- surface reconstruction
- focal length
- polyhedral objects
- camera motion
- reprojection error
- depth information
- feature correspondences
- d objects
- camera pose estimation
- scene structure
- vanishing points
- image based localization
- camera calibration
- perspective camera
- machine learning
- calibrated cameras
- camera positions
- multibody
- smooth surfaces
- three dimensional
- object recognition
- multiple cameras
- field of view