MOSE: Boosting Vision-based Roadside 3D Object Detection with Scene Cues.
Xiahan ChenMingjian ChenSanli TangYi NiuJiang ZhuPublished in: CoRR (2024)
Keyphrases
- d objects
- cad models
- three dimensional
- multi view
- object detectors
- object recognition
- object parts
- cad model
- object classes
- surface patches
- viewpoint
- pose estimation
- real world objects
- multiple views
- geometric models
- object features
- range data
- shape descriptors
- vision system
- surface properties
- d scene
- multiple images
- viewing direction
- object detection
- physically plausible
- image sequences
- computer vision
- pose normalization
- video sequences
- d mesh
- line drawings
- single image
- moving objects
- computer aided design
- object model
- augmented reality
- topological information
- input image
- object views
- integral imaging
- machine learning