saberzl/So-Fake-Set

So-Fake-Set是一个针对社交媒体图像伪造检测的大规模多样化数据集，包含真实图像、全合成图像以及篡改图像。数据集提供了二进制掩码来突出篡改图像中的操作区域，以及用于分类的标签、生成器模型信息、原始文件名和数据集划分信息。该数据集旨在支持社交媒体图像伪造的检测研究。So-Fake-Set is a large-scale, diverse dataset tailored for social media image forgery detection, including real, full_synthetic, and tampered images. The dataset provides binary masks to highlight manipulated regions in tampered images, along with classification labels, generator/source model information, original filenames, and dataset split information. It is designed to support research in the detection of image forgeries on social media.