Real Face vs 3D Resin Mask Anti-Spoofing Dataset

The Real Face vs 3D Resin Mask Anti-Spoofing Dataset is a specialized collection designed to enhance facial recognition systems' resistance to advanced spoofing attacks. Developed by Axon Labs, it comprises over 2,000 high-resolution videos featuring both genuine faces and 3D resin mask attacks, captured from six distinct individuals. These videos are tailored to meet iBeta Level 2 certification requirements, focusing on both active and passive liveness detection.
The dataset includes diverse scenarios such as zoom-in, zoom-out, head tilts, and lateral rotations, filmed under various lighting conditions and with different hairstyles and accessories to simulate real-world environments. This variability ensures comprehensive training for anti-spoofing models, preparing them for real-world deployment.
By providing dual recordings—authentic face videos and corresponding mask attack videos—the dataset offers a robust foundation for developing and testing liveness detection algorithms. Its structured approach aids in the creation of more secure biometric systems capable of distinguishing between genuine users and sophisticated spoofing attempts.

真实人脸 vs 3D树脂面具反欺骗数据集（Real Face vs 3D Resin Mask Anti-Spoofing Dataset）是一款专为增强人脸识别系统抵御高级欺骗攻击能力而打造的专用数据集。该数据集由Axon Labs开发，包含超过2000段高分辨率视频，涵盖真实人脸与3D树脂面具欺骗两类场景，采集自6名不同个体。所有视频均符合iBeta二级认证（iBeta Level 2）标准，覆盖主动与被动活体检测（liveness detection）两大场景。

该数据集包含多样化的拍摄动作与场景，包括变焦拉近、变焦推远、头部俯仰及左右旋转，并在多种光照条件下，搭配不同发型与配饰进行拍摄，以还原真实世界的使用环境。这种多样化的设置能够为反欺骗模型提供全面的训练数据，使其更好地适配真实场景的部署需求。

该数据集提供真实人脸视频与对应面具欺骗视频的双轨录制数据，为活体检测算法的开发与测试提供了坚实的基础。其标准化的采集方案有助于构建更安全的生物识别系统，使其能够精准区分真实用户与各类复杂的欺骗攻击行为。