查看文章

With the increasing potential of Deep fakes in the field of computer vision has made many toilsome tasks effortless. In this paper, we will be discussing one such task. We will demonstrate how we can generate a real like images that don’t even exist in the real world. We will be implementing this with the DCGAN (Deep Convolution GAN) algorithm which is an extended network of GAN (Generative Adversarial Network). Although there are other algorithms available such as encoder and decoder DCGAN has demonstrated to be an incredible accomplishment in generating better quality images. Also, we have talked about the conceptual parts of GAN and examined our technique to make a DCGAN model. For training purposes, we will be using the CelebA dataset which consists of more than 200k faces of celebrities.