NTIRE 2023 Challenge on Image Super-Resolution (×4): Methods and Results

This paper reviews the NTIRE 2023 challenge on image super-resolution (×4), focusing on the proposed solutions and results. The task of image super-resolution (SR) is to generate a high-resolution (HR) output from a corresponding low-resolution (LR) input by leveraging prior information from paired LR-HR images. The aim of the challenge is to obtain a network design/solution capable to produce high-quality results with the best performance (e.g., PSNR). We want to explore how high performance we can achieve regardless of computational cost (e.g., model size and FLOPs) and data. The track of the challenge was to measure the restored HR images with the ground truth HR images on DIV2K testing dataset. The ranking of the teams is determined directly by the PSNR value. The challenge has attracted 192 registered participants, where 15 teams made valid submissions. They achieve state-of-the-art performance in single image super-resolution.

1. Introduction

Single image super-resolution (SR) is a field of research that focuses on the recovery of high-resolution (HR) images from their low-resolution (LR) counterparts that have undergone a certain degradation process. The topic has recently garnered significant attention from both the vision and graphics communities, with a surge of interest in the field. There has been a constant growth of related papers, and substantial progress has been made with employing CNN and Transformer [5, 7, 13, 23, 30, 33, 37, 41, 59, 63, 65].