ON THIS PAGE

• Abstract
• 1. Introduction
• 2. Related Work
• Deep Architectures
• Attention and gating mechanisms
• 3. Squeeze-and-Excitation Blocks
• 3.1 Squeeze : Global Information Embedding
• 3.2 Excitation : Adaptive Recalibration
• Excitation
• Scale
• 3.3 Exemplars : SE-Inception and SE-ResNet
• SE-Inception
• SE-ResNet
• 4. Model and Computational Complexity
• 연산량(FLOPs)와 학습시간
• 파라미터 수 증가량
• 5. Implementation
• 학습 데이터 및 증강
• 학습 인프라
• 최적화 설정
• 6. Experiments
• ImageNet Classification
• Scene Classification
• Object Detection
• Analysis and Interpretation
• Reduction ratio r
• The role of Excitation
• 7. Conclusion
• Code review
• 학습 데이터 및 증강
• 학습 인프라
• 최적화 설정
• SE-ResNet-50
• 결과