DEVIANT: Depth EquiVarIAnt NeTwork for Monocular 3D Object Detection.
Abhinav KumarGarrick BrazilEnrique CoronaArmin ParchamiXiaoming LiuPublished in: CoRR (2022)
Keyphrases
- d objects
- pose estimation
- multi view
- viewpoint
- object recognition
- range data
- multiple views
- shape descriptors
- cad models
- three dimensional
- physically plausible
- integral imaging
- depth cues
- geometric models
- depth images
- line drawings
- object detection
- optical flow
- complex networks
- network structure
- d mesh
- depth map
- object features
- partial matching
- image sequences
- shape from shading
- depth information
- rotation invariant
- camera calibration
- image contours
- machine learning