iBeta 1 - 42,280 Liveness Detection Dataset

The iBeta Level 1 Liveness Detection Dataset is a meticulously curated collection of over 42,000 video attacks, designed specifically to meet iBeta requirements for Level 1 certification of liveness detection systems. This dataset serves as a vital benchmark for developing, evaluating, and improving liveness detection algorithms and anti-spoofing technologies.

Key Features of the Dataset

Diverse Attack Scenarios

The dataset includes 7 unique types of spoofing attacks, covering a wide range of real-world challenges:

1. Outline: Printed outlines of facial photos.
2. Outline3D: Portraits attached to a cylinder for 3D simulation.
3. Mask: Photos with cut-out eyes, used in various configurations.
4. Mask3D: Cardboard masks created from multiple layers of portraits.
5. Phone: Spoofing using photos displayed on smartphones.
6. Monitor: Photos displayed on computer screens.
7. Real: Genuine human faces for baseline comparison.

High-Quality Video Data

• Videos captured on Apple iPhones and Google Pixels, ensuring high resolution and diverse quality.
• Filmed against varied backgrounds and with accessories like facial hair, scarves, and hats to enhance variability.

Applications in Anti-Spoofing Research

• Ideal for training deep neural networks and biometric systems to distinguish between real and spoofed inputs.
• Focused on identifying unseen spoofing cues, making it suitable for cutting-edge anti-spoofing research.
• Supports advancements in face liveness detection, presentation attack detection, and biometric security systems.