A Survey on Leveraging Pre-trained Generative Adversarial Networks for Image Editing and Restoration

TL;DR

This survey reviews recent advances in utilizing pre-trained large-scale GANs for image editing and restoration, highlighting their training, understanding, and application in various tasks.

Contribution

It provides a comprehensive overview of recent progress in leveraging pre-trained GAN models for image editing and restoration, covering training, understanding, and application aspects.

Findings

Pre-trained GANs enable high-quality image editing and restoration.

Understanding GAN priors improves image generation tasks.

Large-scale GANs have significantly advanced image synthesis quality.

Abstract

Generative adversarial networks (GANs) have drawn enormous attention due to the simple yet effective training mechanism and superior image generation quality. With the ability to generate photo-realistic high-resolution (e.g., $1024\times1024$) images, recent GAN models have greatly narrowed the gaps between the generated images and the real ones. Therefore, many recent works show emerging interest to take advantage of pre-trained GAN models by exploiting the well-disentangled latent space and the learned GAN priors. In this paper, we briefly review recent progress on leveraging pre-trained large-scale GAN models from three aspects, i.e., 1) the training of large-scale generative adversarial networks, 2) exploring and understanding the pre-trained GAN models, and 3) leveraging these models for subsequent tasks like image restoration and editing. More information about relevant methods and repositories can be found at https://github.com/csmliu/pretrained-GANs.