KD loss: Enhancing discriminability of features with kernel trick for object detection in VHR remote sensing images

Keyphrases

• Kernel Trick 100%
• Object Detection 100%
• High-resolution Remote Sensing Images (HRRSIs) 100%
• Discriminative Loss 100%
• Kernel-based 100%
• Loss Function 50%
• Class Center 50%
• Feature Representation 33%
• Copyright 16%
• Popular 16%
• Ability to Learn 16%
• Classification Accuracy 16%
• Accurate Classification 16%
• Feature Space 16%
• Nonlinear Relationship 16%
• Softmax 16%
• Discriminative Features 16%
• Kernel Function 16%
• New Loss Function 16%
• Loss Map 16%
• RS Data 16%
• Feature Discriminability 16%
• Cross-entropy Loss 16%
• Hilbert Space 16%
• Mercer Kernel 16%
• High-resolution Remote Sensing 16%
• DOTA Dataset 16%
• Intra-class Distance 16%
• Remote Detection 16%
• Higher-dimensional Hilbert Space 16%
• Class-specific Features 16%
• Inter-class Distance 16%

Computer Science

• Object Detection 100%
• Discriminability 100%
• high resolution remote sensing image 100%
• Deep Feature 66%
• Feature Space 33%
• Hilbert Space 33%
• Classification Accuracy 33%
• Discriminative Feature 33%
• Kernel Function 33%
• Dimensional Hilbert Space 33%
• high resolution remote sensing 33%