LeOCLR: Leveraging Original Images for Contrastive Learning of Visual Representations

Keyphrases

• Visual Representation 100%
• Original Image 100%
• Contrastive Learning 100%
• Representation Learning 66%
• Instance Discrimination 33%
• Object Detection 33%
• Transfer Learning 16%
• Semantic Content 16%
• Supervised Learning 16%
• Loss Function 16%
• Moco 16%
• Semantic Features 16%
• Data Augmentation 16%
• Resizing 16%
• Classification Detection 16%
• Discrimination Method 16%
• Downstream Task 16%
• Augmentation Techniques 16%
• Suboptimal Outcome 16%
• Detection Task 16%
• Object Parts 16%
• ImageNet 16%
• Image Object 16%
• Image Classification 16%
• Random Cropping 16%

Computer Science

• Contrastive Learning 100%
• Visual Representation 100%
• Representation Learning 66%
• Object Detection 33%
• Semantic Feature 16%
• Image Classification 16%
• Supervised Learning 16%
• Baseline Model 16%
• Data Augmentation 16%
• Transfer Learning 16%