Summary: | An investigation into the baseline GAN and progressive GAN (PGGAN) and subsequent works like the style-based GAN architectures (StyleGAN & StyleGAN2) for facial feature disentanglement. Analysis of the structure of the latent space and random distribution will lead to an understanding of the image generation process. In addition, high-level features such as background and foreground, and fine-grained details such as the features of generated images will be discussed. Exploration of various feature disentanglement structures will be done for understanding. Ultimately, a feature disentangling structure based on representation learning architectures will be proposed.
|