3D paper mask attack dataset for Liveness

The 3D Volume-Based Paper Attack Dataset is a cutting-edge resource designed to enhance liveness detection and Presentation Attack Detection (PAD) systems. Focused on 3D volume-based elements such as noses, shoulders, and foreheads, it targets advanced paper attack scenarios. This dataset is invaluable for training and evaluating AI models for biometric authentication and anti-spoofing technology.

Key Features

• Diverse Participants: Over 40 participants representing a balanced mix of genders and ethnicities, including Caucasian, Black, and Asian groups.
• Rich Data Collection:
  • Captured using both iOS and Android devices.
  • Includes multiple frames from ~7-second videos with zoom-in and zoom-out phases for active liveness testing.
  • Variety of paper types and specific attack scenarios with movement.
• Customizable Options:
  • Add volume-based elements like scarves, glasses, and hoodies.
  • Data collected from low-end to high-end capturing devices.
• Advanced Applications: Supports PAD Level 1 and Level 2 liveness tests, making it suitable for tackling the most sophisticated spoofing attempts.

Applications

• Liveness Detection: Train AI models to distinguish between real selfies and spoof attacks with exceptional accuracy.
• iBeta Certification Preparation: Valuable for developing and testing liveness detection models before applying for iBeta certification.
• Biometric Security: Enhance facial recognition systems for biometric authentication and fraud prevention.