MatchU: Matching Unseen Objects for 6D Pose Estimation from RGB-D Images

Keyphrases

• 2D Texture 25%
• 2D to 3D 25%
• 3D Descriptor 25%
• 3D Geometric 25%
• 3D Geometry 25%
• 6D Pose 25%
• 6D Pose Estimation 100%
• Agnostic 25%
• Attention Mechanism 25%
• By Design 25%
• Cross-Modal Data 25%
• Descriptor Learning 25%
• Fusion Strategy 25%
• Generic Approach 25%
• Geometric Cues 25%
• Individual Objects 25%
• Intensive Training 25%
• Latent Space 25%
• Learned Descriptors 25%
• Learning Methods 25%
• Learning Process 25%
• Match Strategy 25%
• Matching Losses 25%
• Novel Design 25%
• Object Categories 25%
• Object Instance 25%
• Object Pose Estimation 25%
• Pose Prediction 25%
• Real Application 25%
• RGB-D 100%
• RGB-D Data 25%
• Rotation Invariance 25%
• Unseen Objects 100%

Computer Science

• Generalizability 33%
• Individual Object 33%
• Learning Process 33%
• Object Category 33%
• Object Instance 33%
• Object Pose 33%
• Pose Estimation 100%
• Real Application 33%
• RGB-D Image 100%
• Rotation Invariant 33%