2D Masks with Eyeholes Attacks - 11,200+ videos

Dataset Overview

The 2D Masks with Eyeholes Attacks Dataset includes over 11,200 videos of individuals wearing or holding 2D printed masks with eyeholes. These videos were captured using 5 different devices, making the dataset a valuable resource for studying presentation attacks in biometric security systems. Specifically, it targets spoofing attacks on facial recognition technologies.

This extensive collection is designed to support the development and refinement of liveness detection and spoofing detection techniques, which are critical for meeting iBeta Level 1 & 2 certification requirements for facial recognition systems.

Dataset Overview

• Total Videos: 11,200+
• Devices Used: 5 different camera devices
• Focus Area: Presentation attacks and spoofing detection in facial recognition systems
• Target: Liveness detection and spoofing prevention for biometric security systems

Key Features

• Realistic Mask Attacks: The dataset includes videos of people wearing or holding 2D printed masks with eyeholes, simulating spoofing attempts.
• Diverse Devices: Videos were captured using 5 different devices, ensuring variation in angles, lighting, and camera quality. This diversity is crucial for conducting comprehensive research.
• Advanced Spoofing Detection: Essential for developing and testing liveness detection algorithms to protect against spoofing attacks on facial recognition systems.

Applications

• Facial Recognition Security: This dataset helps train models to detect spoofing attempts using 2D masks, improving the security of facial recognition systems.
• Liveness Detection: Enhance liveness detection techniques, ensuring the authenticity of facial recognition in security systems.
• Biometric Research: Use the dataset for studies on biometric security, spoofing prevention, and to meet certification standards like iBeta Level 1 & 2.