SfM-TTR: Using Structure from Motion for Test-Time Refinement of Single-View Depth Networks.
Sergio IzquierdoJavier CiveraPublished in: CoRR (2022)
Keyphrases
- structure from motion
- single view
- multi view
- bundle adjustment
- depth map
- point cloud
- surface reconstruction
- perspective projection
- missing data
- camera parameters
- image sequences
- camera pose
- multiple views
- computer vision
- scene depth
- motion stereo
- polyhedral objects
- focal length
- motion estimates
- camera motion
- feature tracking
- feature correspondences
- depth information
- vanishing points
- scene structure
- reprojection error
- image based localization
- light field
- articulated objects
- camera positions
- training set
- camera calibration
- high resolution
- d objects
- range images
- multiple cameras
- stereo vision