GANDiffFace

GANDiffFace是由马德里自治大学生物识别和数据模式分析实验室开发的一个用于人脸识别的合成数据集。该数据集包含700个合成身份，每个身份有多张图像，涵盖了多种配件、姿势、表情和场景，以增强类内变化。数据集的创建结合了生成对抗网络（GAN）和扩散模型的优势，旨在解决现有合成数据集在类内变化有限、缺乏真实感和不公平地代表人口统计群体等问题。GANDiffFace数据集旨在为训练和测试多种面部分析应用提供一个具有可控和真实类内变化的数据集，同时确保人口统计群体的公平代表性。

GANDiffFace is a synthetic face recognition dataset developed by the Biometrics and Data Pattern Analysis Laboratory of the Autonomous University of Madrid. This dataset contains 700 synthetic identities, each with multiple images covering a diverse set of accessories, poses, expressions and scenes to augment intra-class variation. Developed by combining the strengths of generative adversarial networks (GANs) and diffusion models, the dataset is designed to address the shortcomings of existing synthetic datasets, including limited intra-class variation, poor realism, and unfair representation of demographic groups. The GANDiffFace dataset aims to provide a dataset with controllable and realistic intra-class variation for training and testing various facial analysis applications, while ensuring fair representation across demographic groups.