Anti-Spoofing Dataset: Cutout 2D Mask Attacks

The "Anti-Spoofing Dataset: Cutout 2D Mask Attacks" is designed to aid in developing AI models for detecting cutout 2D mask spoofing attacks in facial liveness detection systems. This dataset comprises over 2,000 participants, offering a diverse range of facial images and videos captured under various conditions. The data includes both genuine and spoofed samples, with the spoofed instances featuring flat printed masks with cut-out eye and mouth holes, simulating real faces. These masks are typically used to deceive biometric authentication systems. The dataset is balanced in terms of gender and ethnicity, enhancing its applicability for training robust anti-spoofing models. It provides a valuable resource for researchers and developers working on improving the security and reliability of facial recognition technologies.

“防欺骗数据集：抠图2D面具攻击”（Anti-Spoofing Dataset: Cutout 2D Mask Attacks）旨在助力开发用于人脸活体检测系统中抠图2D面具欺骗攻击检测的AI模型。该数据集包含超过2000名参与者的样本，提供了在多种采集条件下获取的多样化人脸图像与视频数据。数据涵盖真实人脸样本与欺骗攻击样本两类，其中欺骗攻击样本采用带有抠出眼部与嘴部孔洞的平面印刷面具以模拟真实人脸，此类面具常被用于欺骗生物特征认证系统。该数据集在性别与种族分布上保持均衡，可提升训练鲁棒防欺骗模型的适用性，为致力于提升人脸识别技术安全性与可靠性的研究人员与开发者提供了宝贵的研究资源。