3D-RCNN: Instance-Level 3D Object Reconstruction via Render-and-Compare

Keyphrases

• 3D Shape 100%
• Instance-level 100%
• 3D Pose 100%
• 3D Object Reconstruction 100%
• Render-and-compare 100%
• Depth Map 50%
• Class-specific 50%
• Image Region 50%
• Novel Representation 50%
• Object Instance 50%
• Equivariance 50%
• 3D Scene Modeling 50%
• Shape Prior 50%
• CAD Model 50%
• 3D Scene Understanding 50%
• 3D Representation 50%
• Shape Space 50%
• Autonomous Driving 50%
• Deep Convolutional Network 50%
• Instance Segmentation 50%
• Map Image 50%
• Supervisory Signal 50%
• Segmentation Map 50%
• KITTI 50%
• Graphics Framework 50%
• 2D Vision 50%
• Inverse Graphics 50%
• Differentiable Rendering 50%

Computer Science

• Object Reconstruction 100%
• 3d Scenes 100%
• Instance Level 100%
• Annotation 50%
• Computer Aided Design 50%
• Scene Understanding 50%
• Object Instance 50%
• Convolutional Network 50%
• Instance Segmentation 50%
• Autonomous Driving 50%