Consistent 3D Hand Reconstruction in Video via Self-Supervised Learning

Keyphrases

• Self-supervised Learning 100%
• 3D Hand Reconstruction 100%
• Training Data 50%
• Benchmark Dataset 50%
• Video Data 50%
• Motion Information 50%
• Reconstruction Model 50%
• Reconstruction Accuracy 50%
• Supervised Methods 50%
• Reconstruction Performance 50%
• Motion Consistency 50%
• Jointly Estimate 50%
• Image Texture 50%
• Hand Motion 50%
• Hand Pose 50%
• Hand Reconstruction 50%
• Self-supervised Method 50%
• Camera Viewpoint 50%
• Monocular Video 50%
• Video Training 50%
• Hand Keypoints 50%
• Motion Texture 50%
• Texture Consistency 50%
• Shape Consistency 50%

Computer Science

• Supervised Method 100%
• Self-Supervised Learning 100%
• Training Data 50%
• Motion Information 50%
• Annotation 50%
• Reconstruction Accuracy 50%
• Video Training 50%