AdvIT: Adversarial frames identifier based on temporal consistency in videos

Keyphrases

• Action Recognition 50%
• Adaptive Attack 50%
• Adaptive Attackers 50%
• Adversarial Examples 50%
• Adversary 50%
• Attack Strategy 50%
• Attacker 50%
• Autonomous Driving System 50%
• Autonomous Surveillance 50%
• Consistency Properties 50%
• Critical Security 50%
• Deep Neural Network 100%
• Defense Approaches 50%
• Detection Method 50%
• Detection Rate 100%
• Effective Method 50%
• Efficient Processing 50%
• Frame Detection 50%
• Human Pose Estimation 50%
• Image Classification 50%
• Inconsistency 50%
• Learning Task 50%
• Object Detection & Recognition 50%
• Optical Flow 50%
• Optical Flow Estimation 50%
• Real-time Detection 50%
• Static Image 50%
• Surveillance System 50%
• Temporal Consistency 100%
• Transferability 50%
• Transformation Method 50%
• Video Analysis 50%
• Video Semantic Segmentation 50%

Computer Science

• Action Recognition 50%
• Adversarial Example 50%
• Attackers 100%
• Autonomous Driving 50%
• Consistency Property 50%
• Deep Neural Network 100%
• Detection Rate 100%
• Effective Method 50%
• Image Classification 50%
• Image Segmentation 50%
• Object Detection 50%
• Optical Flow Estimation 50%
• Pose Estimation 50%