Self-supervised Wide Baseline Visual Servoing via 3D Equivariance

Keyphrases

• Visual Servoing 100%
• Equivariance 100%
• Geometric Properties 40%
• Closed-loop 20%
• Occlusion 20%
• Success Rate 20%
• Visual Representation 20%
• Drastic Change 20%
• Error Reduction 20%
• Camera Views 20%
• Feature Space 20%
• Geodesics 20%
• Average Distance 20%
• Learning Model 20%
• Relative Transformation 20%
• Error Tolerance 20%
• Object Appearance 20%
• 3D Transformation 20%
• Distance Error 20%
• Ground Truth Data 20%
• Outperformance 20%
• Wide-baseline Images 20%
• Object Alignment 20%
• Geodesic Space 20%
• Absolute Camera Pose 20%
• 3D Bounding Box 20%
• Siamese Network 20%
• Task Alignment 20%

Computer Science

• Geodesics 100%
• Feature Space 50%
• Visual Representation 50%
• Siamese Neural Network 50%
• Error Tolerance 50%
• Error Distance 50%