How PARTs assemble into wholes: Learning the relative composition of images

Keyphrases

• Self-supervised Learning 100%
• Pretext Task 100%
• Relative Transformation 50%
• Relative Positioning 50%
• Global Classifier 50%
• Fixed Grid 50%
• Partial Visibility 50%
• Positional Relationship 50%
• Object Composition 50%
• Location Index 50%
• Spatial Understanding 50%
• Grid-based Approach 50%
• Spatial-aware 50%
• Grid Structure 50%
• Grid-based Method 50%
• Universal Self 50%
• Grid Constraints 50%
• Absolute Position 50%
• Stylistic Change 50%

Computer Science

• Self-Supervised Learning 100%
• Positional Relationship 50%
• Relative Positioning 50%
• Spatial Understanding 50%

Arts and Humanities

• Self-Supervised Learning 66%
• Electroencephalography 33%
• stylistic changes 33%