Improving the quality of facial attribute transfer images based on generative adversarial networks

Recently, facial attribute transfer has become an important research direction in the fields of image processing, but how to improve the quality of the final synthesized faces has always been a difficult point for continuous improvement. Therefore, an improved face attribute transfer model based on generative adversarial networks (GAN) is proposed in this paper. The PatchGAN structure was introduced to train the network model, and the discriminator structure was entirely composed of convolutional neural networks, encouraging the model to give greater emphasis to the details of the generated images. The experimental results show that the proposed improvement method significantly improves the details of generated images and enhances the quality of generated images.