3DQ-Nets: Visual Concepts Emerge in Pose Equivariant 3D Quantized Neural Scene Representations.
Mihir PrabhudesaiShamit LalHsiao-Yu Fish TungAdam W. HarleyShubhankar PotdarKaterina FragkiadakiPublished in: CVPR Workshops (2020)
Keyphrases
- visual concepts
- visual data
- image content
- video sequences
- learning tasks
- video content
- pose estimation
- image collections
- visual information
- visual content
- semantic concepts
- object categories
- d objects
- image sequences
- semantic gap
- training data
- visual features
- positive examples
- spatial relations
- image data
- text categorization
- image annotation
- image set
- object detection
- multi modal
- higher level
- video data