Realization of Virtual Human Face Based on Deep Convolutional Generative Adversarial Networks

摘要

At present, the generative adversarial networks research that generates a high confidence image for a large number of training samples has achieved some results, but the existing research only performs image generation for known training samples, but does not use the training parameters for image generation other than training samples. This paper uses the TensorFlow deep learning framework to build deep convolutional generative adversarial networks to complete the generation of virtual face images. From the experimental results, it can better generate virtual face images similar to real faces, which provides new ideas and methods for the research of generating virtual images.