Generation of High-Quality Realistic Faces with StyleGAN
K. Ashwini, D. Nagajyothi, Ch. Ramakrishna, V. Jyothi · 2023
The quality of images generated using traditional GANs for face image generation is limited because the generator and discriminator share the same backpropagation network. This paper presents algorithms that aim to enhance the quality of generated images, specifically in the context of high-quality face image generation. In this paper, we use techniques like styleGAN which trump over techniques like DCGAN in face generation. We also generate a video of the morphing faces showing the full capabilities of styleGAN. This paper goes into the details of GANs, StyleGANs, and DCGANs and tries to compare the results of various GANs in doing the task above. The dataset utilized is called FFHQ.