Efficiency Evaluation of Popular Deepfake Methods Using Convolution Neural Network

Many deepfake techniques in the early years are spread to create successful videos (i.e., Face Swap, Deep Fake, etc.). These methods enable anyone manipulate faces videos, which can negatively impact society. One way reduce this problem is detection. It has become such a hot topic and most crucial task recent years. This paper proposes deep learning model detect evaluate video using convolutional neural networks. The evaluated on FaceForensics++ dataset that contains four different ways (deepfake, face 2 face, swap, neuraltexture), it achieved 0.96 accuracy method, 0.95 approach, 0.94 precision swap method 0.76 neuraltexture method.