Anti Spoofing Real Dataset - 5,000+ files

The Anti Spoofing Real Dataset – 5,000+ files, hosted on Kaggle by Axon Labs, is a live-face biometric dataset designed for spoof-detection research. It contains over 5,000 genuine selfie images captured under real-world conditions. The dataset spans a wide demographic diversity, including participants from more than 50 countries, providing global representation and varied facial attributes Kaggle.Each file corresponds to an authentic self-portrait taken with different mobile devices, encompassing a variety of lightings, head poses, and environmental contexts. These real-life selfies serve as a valuable baseline for training and benchmarking liveness detection and presentation attack detection (PAD) systems, allowing researchers to distinguish genuine faces from spoofed attempts effectively Kaggle. Overall, the collection provides a robust foundation for improving anti-spoofing models by emphasizing diversity and real-world variability. Its size and global scope make it particularly useful for evaluating algorithm performance and generalization across diverse user populations.

由Axon Labs上传至Kaggle的反欺骗真实数据集（Anti Spoofing Real Dataset）包含5000余个文件，是一款专为欺骗检测研究打造的活体生物特征数据集。该数据集收录了5000余张在真实场景下拍摄的真实自拍图像，涵盖广泛的人口统计学多样性，参与者来自50余个国家，具备全球代表性且面部属性丰富多样。每个文件对应一张使用不同移动设备拍摄的真实自拍照，涵盖多样的光照条件、头部姿态与环境背景。这些真实场景下的自拍可作为训练与基准测试活体检测及呈现攻击检测（Presentation Attack Detection, PAD）系统的宝贵基准，助力研究人员有效区分真实人脸与欺骗伪造的人脸尝试。总体而言，该数据集通过强调多样性与真实场景变异性，为优化反欺骗模型提供了坚实的基础。其规模与全球覆盖范围使其特别适用于评估算法在多样化用户群体中的性能与泛化能力。