HydraFake-100K
收藏资源简介:
HydraFake-100K数据集是中国科学院自动化研究所、蚂蚁集团、中国科学院深圳先进技术研究院和中国科学院大学人工智能学院联合创建的。该数据集包含10万条数据,旨在模拟现实世界中的挑战,并通过分层泛化测试推进深度伪造检测的评价协议。数据集收集了来自8个公共数据集的真实图像,以及来自经典数据集、高质量公共数据集和自构建深度伪造数据的伪造图像。HydraFake数据集还包含了来自社交网络的1千张深度伪造图像,以模拟真实世界中的挑战。
The HydraFake-100K dataset was jointly created by the Institute of Automation of the Chinese Academy of Sciences, Ant Group, Shenzhen Institute of Advanced Technology of the Chinese Academy of Sciences, and the School of Artificial Intelligence of the University of Chinese Academy of Sciences. Comprising 100,000 data samples, this dataset is designed to simulate real-world challenges and advance the evaluation protocols for deepfake detection via hierarchical generalization testing. It collects real images from 8 public datasets, as well as forged images sourced from classic datasets, high-quality public datasets, and self-constructed deepfake data. Additionally, the HydraFake dataset includes 1,000 deepfake images from social networks to simulate real-world challenges.




