CelebA-Spoof 具有丰富注释的大规模人脸反欺骗数据集

CelebA-Spoof 是一个大规模的人脸反欺骗数据集，包含来自 10,177 个主题的 625,537 张图像，其中包括人脸、光照、环境和欺骗类型的 43 个丰富属性。从 CelebA 数据集中选择的实时图像。我们收集并注释 CelebA-Spoof 的恶搞图像。在 43 个丰富的属性中，40 个属性属于 Live 图像，包括所有面部组件和配件，如皮肤、鼻子、眼睛、眉毛、嘴唇、头发、帽子、眼镜。 3个属性属于恶搞图像，包括恶搞类型、环境和光照条件。CelebA-Spoof可用于训练和评估人脸反欺骗算法。

CelebA-Spoof is a large-scale face anti-spoofing dataset containing 625,537 images from 10,177 subjects, with 43 rich attributes spanning face traits, illumination, environment and spoof categories. Live images are selected from the CelebA dataset, while we collected and annotated the spoof images for the CelebA-Spoof dataset. Out of the 43 attributes, 40 are for live images, covering all facial components and accessories such as skin, nose, eyes, eyebrows, lips, hair, hats and glasses. The remaining 3 attributes are for spoof images, including spoof type, environment and illumination conditions. CelebA-Spoof can be utilized for training and evaluating face anti-spoofing algorithms.

搜集汇总

背景与挑战

背景概述

CelebA-Spoof是一个包含625,537张图像的大规模人脸反欺骗数据集，涵盖10,177个主题，并提供了43个丰富属性注释，包括人脸、光照、环境和欺骗类型等。该数据集可用于训练和评估人脸反欺骗算法。