MV-FCOS3D++: Multi-View Camera-Only 4D Object Detection with Pretrained Monocular Backbones.
Tai WangQing LianChenming ZhuXinge ZhuWenwei ZhangPublished in: CoRR (2022)
Keyphrases
- multi view
- d objects
- multiple cameras
- camera calibration
- pose estimation
- camera parameters
- multiple views
- single view
- depth map
- viewpoint
- stereo camera
- three dimensional
- view invariant
- multiple viewpoints
- dynamic scenes
- focal length
- range data
- position and orientation
- photometric stereo
- field of view
- object recognition
- viewing direction
- surface reconstruction
- ego motion
- bundle adjustment
- image sequences
- multi camera
- object classes
- view synthesis
- depth information
- ground plane
- vision system
- multi views
- d mesh
- vanishing points
- visual hull
- camera pose
- single image