DeepFake Detection Challenge (DFDC) Dataset
收藏资源简介:
DeepFake Detection Challenge (DFDC) Dataset是由Facebook AI创建的一个大规模人脸交换视频数据集,旨在训练深度伪造检测模型。该数据集包含超过100,000个视频片段,来源于3,426名付费演员,使用多种深度伪造、GAN基和非学习方法制作。数据集的创建过程涉及确保所有参与者同意其肖像被修改,并记录在多种自然环境下。DFDC数据集的应用领域主要集中在解决深度伪造视频的检测问题,旨在通过大规模数据训练提高检测模型的泛化能力。
DeepFake Detection Challenge (DFDC) Dataset is a large-scale face-swapping video dataset developed by Facebook AI, designed to train deepfake detection models. This dataset contains over 100,000 video clips sourced from 3,426 paid actors, and was created using multiple deepfake, GAN-based, and non-learning-based methods. The development of the dataset ensured that all participants consented to their likeness being modified, and the videos were recorded in various natural environments. The main application of the DFDC dataset focuses on addressing deepfake video detection, with the goal of improving the generalization ability of detection models through large-scale data training.




