Face Forgery in the Semantic Context (FFSC)

FFSC数据集由香港城市大学计算机科学系创建，包含147184条数据，旨在通过语义上下文定义面部伪造。数据集中的每张图像都与一个层次图组织的一系列标签相关联，支持两种新的测试协议，以探究面部伪造检测器的泛化能力。此外，FFSC数据集还引入了一种面向语义的面部伪造检测方法，该方法通过捕捉标签关系并优先处理主要任务（即真实或伪造面部检测）来提高检测性能。数据集的应用领域主要集中在提高面部伪造检测的准确性和泛化性，解决当前检测器依赖特定伪造场景特征的问题。

The FFSC dataset, developed by the Department of Computer Science, City University of Hong Kong, contains 147,184 data samples, and is constructed to define facial forgery through semantic context. Each image in the dataset is linked to a set of labels organized in a hierarchical graph structure, which supports two novel test protocols for exploring the generalization capability of facial forgery detectors. Additionally, the FFSC dataset introduces a semantic-oriented facial forgery detection method that enhances detection performance by capturing label relationships and prioritizing the core task, namely real or forged facial detection. The primary applications of this dataset center on improving the accuracy and generalization of facial forgery detection, aiming to resolve the limitation that current detectors heavily rely on features specific to individual forgery scenarios.