Self-Supervised Face Image Restoration with a One-Shot Reference

TL;DR

This paper introduces SAIR, a semantic-aware method for face image restoration that leverages reference images to improve quality and semantic accuracy, outperforming existing approaches.

Contribution

SAIR explicitly models semantic information from reference images, enabling reliable restoration and semantic correction of severely degraded face images.

Findings

SAIR achieves superior restoration quality in experiments.

SAIR effectively corrects semantic inconsistencies.

Quantitative results show improved metrics over baseline methods.

Abstract

For image restoration, methods leveraging priors from generative models have been proposed and demonstrated a promising capacity to robustly restore photorealistic and high-quality results. However, these methods are susceptible to semantic ambiguity, particularly with images that have obviously correct semantics such as facial images. In this paper, we propose a semantic-aware latent space exploration method for image restoration (SAIR). By explicitly modeling semantics information from a given reference image, SAIR is able to reliably restore severely degraded images not only to high-resolution and highly realistic looks but also to correct semantics. Quantitative and qualitative experiments collectively demonstrate the superior performance of the proposed SAIR. Our code is available at https://github.com/Liamkuo/SAIR.