AxonData/Anti_spoofing_dataset_video_replay_attack

Liveness Detection Dataset包含真实自拍和超过5000个基于这些自拍的重放攻击。数据集旨在帮助训练和评估活体检测模型，以区分真实自拍和重放攻击。真实自拍来自1000多个个体，质量至少为720p，且面部清晰无滤镜。重放攻击在光照、设备和屏幕方面具有多样性，视频长度至少为12秒，展示了不同角度的攻击。数据集适用于活体检测、反欺诈、生物识别认证、机器学习和深度学习等领域。

The Liveness Detection Dataset contains selfies of real people along with over 5,000 replay display attacks generated from these selfies. The dataset is designed to aid in the training and evaluation of liveness detection models, which distinguish between real selfies and replay attacks. The real life selfies include clear, unfiltered face photos at least 720p quality, while the replay display attacks include attacks with varied lighting, devices, and screens, lasting at least 12 seconds and showing attacks from various angles. Key features include over 1,000 individuals selfies balanced in terms of gender and ethnicity, and more than 5,000 replay display attacks. The dataset is ideal for training and evaluating models for liveness detection, anti-spoofing, biometric authentication, and more.