The development of powerful image generation models has accelerated the development of face manipulation techniques. Deep Fake recognition for Expression Reenactment and Phasing Face Recognition attract much attention for their flexibility and ease of generating high-quality synthesis results. Recently, both subjects have been the subject of active research. However, most existing methods ignore the similarities between the two tasks. Face recognition is significantly impacted by Deepfake recognition. This paper develops a state-of-the-art method for classifying and detecting deepfake and deepswap. Evaluation results demonstrate the generalization capability and performance of the algorithm on different deepfake datasets with extensive experiments on the challenging FaceForensic+ + datasets.