Shift R-CNN: Deep Monocular 3D Object Detection with Closed-Form Geometric Constraints.
Andretti NaidenVlad PaunescuGyeongmo KimByeongMoon JeonMarius LeordeanuPublished in: CoRR (2019)
Keyphrases
- closed form
- d objects
- geometric constraints
- pose estimation
- viewpoint
- object matching
- epipolar geometry
- point correspondences
- multiple views
- multi view
- camera calibration
- object recognition
- visual hull
- three dimensional
- multiple images
- planar surfaces
- shape descriptors
- range data
- object classes
- image sequences
- closed form solutions
- single view
- ground plane
- image pairs