Deepfake Face Generation Techniques Based on Generative Adversarial Networks

With the continuous rise of deepfake technology in recent years, deepfake face generation technology has gradually become a popular technology in the field of computer vision. At the same time, the development of deep learning provides new ideas and solutions for deepfake face generation technology. Generative Adversarial Networks (GAN), as one of the key technologies of current deepfake, has a series of derivative models that make the face images generated by deepfake more realistic. This paper discusses the development of GAN-based deepfake face generation technology. First, the basic principles of deepfake and GAN are introduced respectively. Secondly, the GAN models for deepfake face generation are divided into three categories: early classical models, models that can generate specified attributes, and models guided by text, and the representative GAN models in each category and their advantages and disadvantages are introduced in chronological order. Finally, the current status and future development of deepfake face generation technology and GAN are summarized and outlooked.