Deepfake Detection Challenge (DFDC) Preview Dataset
收藏资源简介:
Deepfake Detection Challenge (DFDC) Preview Dataset是由Facebook AI创建的一个包含5000多条视频的数据集,旨在促进深度伪造检测技术的研究。该数据集通过与演员签订协议,确保了数据使用的合法性,并考虑了性别、肤色、年龄等多维度的多样性。数据集中的视频内容丰富,包括不同背景和光照条件下的面部交换视频。创建过程中,采用了两种面部交换算法,确保了数据的真实性和挑战性。该数据集主要应用于检测和防止恶意使用的深度伪造视频,如用于骚扰、误导公众等,对于维护网络安全和社会稳定具有重要意义。
Deepfake Detection Challenge (DFDC) Preview Dataset is a dataset containing over 5,000 videos created by Facebook AI, aimed at advancing research in deepfake detection technologies. The dataset ensures the legality of data usage by signing agreements with participating actors, and incorporates multi-dimensional diversity covering gender, skin color, age and other attributes. The videos in the dataset feature diverse content, including face-swapped videos under various backgrounds and lighting conditions. Two face-swapping algorithms were adopted during the dataset's creation to guarantee its authenticity and challenging nature. This dataset is primarily applied to detect and prevent malicious deepfake videos, such as those used for harassment or public misinformation, and is of great significance for safeguarding cybersecurity and social stability.

- 1The Deepfake Detection Challenge (DFDC) Preview DatasetFacebook AI · 2019年



