Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Longarela, Bruno, Conde, Marcos V., Garcia, Alvaro, Timofte, Radu
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2510.12765
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866914092386615296
author Longarela, Bruno
Conde, Marcos V.
Garcia, Alvaro
Timofte, Radu
author_facet Longarela, Bruno
Conde, Marcos V.
Garcia, Alvaro
Timofte, Radu
contents This paper presents a comprehensive study and benchmark on Efficient Perceptual Super-Resolution (EPSR). While significant progress has been made in efficient PSNR-oriented super resolution, approaches focusing on perceptual quality metrics remain relatively inefficient. Motivated by this gap, we aim to replicate or improve the perceptual results of Real-ESRGAN while meeting strict efficiency constraints: a maximum of 5M parameters and 2000 GFLOPs, calculated for an input size of 960x540 pixels. The proposed solutions were evaluated on a novel dataset consisting of 500 test images of 4K resolution, each degraded using multiple degradation types, without providing the original high-quality counterparts. This design aims to reflect realistic deployment conditions and serves as a diverse and challenging benchmark. The top-performing approach manages to outperform Real-ESRGAN across all benchmark datasets, demonstrating the potential of efficient methods in the perceptual domain. This paper establishes the modern baselines for efficient perceptual super resolution.
format Preprint
id arxiv_https___arxiv_org_abs_2510_12765
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Efficient Perceptual Image Super Resolution: AIM 2025 Study and Benchmark
Longarela, Bruno
Conde, Marcos V.
Garcia, Alvaro
Timofte, Radu
Computer Vision and Pattern Recognition
This paper presents a comprehensive study and benchmark on Efficient Perceptual Super-Resolution (EPSR). While significant progress has been made in efficient PSNR-oriented super resolution, approaches focusing on perceptual quality metrics remain relatively inefficient. Motivated by this gap, we aim to replicate or improve the perceptual results of Real-ESRGAN while meeting strict efficiency constraints: a maximum of 5M parameters and 2000 GFLOPs, calculated for an input size of 960x540 pixels. The proposed solutions were evaluated on a novel dataset consisting of 500 test images of 4K resolution, each degraded using multiple degradation types, without providing the original high-quality counterparts. This design aims to reflect realistic deployment conditions and serves as a diverse and challenging benchmark. The top-performing approach manages to outperform Real-ESRGAN across all benchmark datasets, demonstrating the potential of efficient methods in the perceptual domain. This paper establishes the modern baselines for efficient perceptual super resolution.
title Efficient Perceptual Image Super Resolution: AIM 2025 Study and Benchmark
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2510.12765