基于三维人脸数据增强的深度伪造检测方法

Deepfake detection based on 3D face data augmentation

随着深度伪造技术的发展,深度伪造视频的制作及传播变得越来越容易,给社会带来了巨大的安全风险,深度伪造检测算法成为当前网络安全领域的重要方向。聚焦于提出一种泛化性能更好、效率更高、可解释性更强的深度伪造检测算法,主要针对DFDC、FaceForensic++及Celeb-DF三个深度伪造视频数据集进行实验并以这三个数据集集中训练出检测模型,首先使用人脸检测算法MTCNN提取人脸图像,进而将EfficientNet网络与Transformer架构相结合作为检测模型,通过采用三维人脸数据增强、注意力机制以及全局局部融合方法对模型进行训练和测试。模型在未使用型集成、知识蒸馏等方法的基础上,达到了与最优检测效果相当的检测水平。

With the development of deepfake technology,the production and dissemination of deepfake videos have become increasingly easy,posing significant security risks to society.Therefore,researching deepfake detection algorithms has become an important direction in the field of network security.This paper focuses on proposing a deepfake detection algorithm with better generalization performance,higher efficiency,and stronger interpretability.Experiments are conducted on three deepfake video datasets:DFDC,FaceForensic++,and Celeb-DF.Firstly,the Multi-task Cascaded Convolutional Networks(MTCNN)face detection algorithm is used to extract facial images.Then,the EfficientNet network is combined with the Transformer architecture as the detection model.The model is trained and tested using data augmentation,attention mechanisms,and global-local fusion methods.Without employing complex model ensembles or knowledge distillation,our model achieves comparable detection performance to state-of-the-art methods.