Anti-Spoofing Dataset: Display Spoof Attack

The Anti-Spoofing Dataset: Display Spoof Attack on Kaggle is designed for liveness detection in biometric systems, focusing specifically on replay-display (screen-based) spoofing. It features high-quality, real selfies contributed by over 1,000 participants, which are subsequently used to generate more than 5,000 video-based display spoofing attacks performed by a team of over 200 individuals.Each genuine selfie is captured at a minimum of 720p resolution, with no filters applied, ensuring clarity and authenticity of facial features. The replay attacks are carefully recorded under diverse conditions—spanning different lighting environments, camera angles, and types of display devices—with each attack video lasting at least 12 seconds, simulating realistic threat scenarios.This dataset supports research and development of robust face liveness detection models capable of discerning genuine faces from sophisticated display-based spoofing attempts. Its balanced representation of demographic diversity (gender and ethnicity) and scenario variety make it a valuable resource for evaluating and improving anti-spoofing technologies.Kaggle平台展示欺骗攻击反欺骗数据集专为生物识别系统（biometric systems）中的活体检测（liveness detection）任务设计，重点聚焦于重放展示（基于屏幕）类欺骗攻击。该数据集收录了超过1000名参与者拍摄的高质量真实自拍照，后续由200余名工作人员基于这些样本生成了超过5000段视频类展示欺骗攻击样本。每一张原始真实自拍照均采用不低于720p分辨率拍摄，且未添加任何滤镜，确保面部特征清晰可辨且真实可信。所有重放攻击样本均在多样化场景下精心录制，涵盖不同光照环境、拍摄角度以及显示设备类型，每段攻击视频时长不少于12秒，用以模拟真实的威胁场景。本数据集可用于研发能够精准区分真实人脸与复杂展示类欺骗攻击的高性能面部活体检测模型。该数据集在人口统计学维度（性别与种族）上分布均衡，且覆盖丰富的场景类型，是评估与改进反欺骗技术的宝贵研究资源。

搜集汇总

背景与挑战

背景概述

该数据集专注于生物识别系统中的活体检测，针对重放显示欺骗攻击，包含来自1000多名参与者的高清真实自拍，以及由200多人执行的超过5000个视频显示欺骗攻击。攻击视频模拟多种光照、角度和显示设备条件，每段至少12秒，可用于开发能区分真脸与复杂显示欺骗的鲁棒活体检测模型。