FS-Net: Fast Shape-based Network for Category-Level 6D Object Pose Estimation with Decoupled Rotation Mechanism

Keyphrases

• Shape-based 100%
• Rotation Mechanism 100%
• 6D Pose Estimation 100%
• Category-level 100%
• Feature Extraction 40%
• 3D Graph Convolution 40%
• Latent Feature 40%
• Point Object 40%
• Training Data 20%
• Benchmark Dataset 20%
• Generalization Ability 20%
• Synthetic Data 20%
• Deformation Mechanism 20%
• Autoencoder 20%
• Size Estimation 20%
• Decoder 20%
• Art Performance 20%
• RGB-D 20%
• Instance-level 20%
• Pose Feature 20%
• 3D Deformation 20%
• Inference Speed 20%
• Category-level Pose Estimation 20%
• Latent Feature Extraction 20%
• Invariance Property 20%
• Point-shift 20%
• Object Size 20%
• Shift Size 20%
• Orientation-aware 20%
• Shift Invariance 20%
• Scale Invariance 20%

Computer Science

• Pose Estimation 100%
• Object Pose 100%
• Feature Extraction 75%
• Graph Convolution 50%
• Training Data 25%
• Synthetic Data 25%
• Art Performance 25%
• Generalization Ability 25%
• Autoencoder 25%
• Instance Level 25%
• Shift Invariance 25%
• RGB-D Image 25%
• Invariance Property 25%
• Scale Invariance 25%

Engineering

• Pose Estimation 100%
• Feature Extraction 75%
• Ground Truth 50%
• Level Feature 25%
• Deformation Mechanism 25%
• Autoencoder 25%
• Object Point 25%
• RGB-D Image 25%