List your company
ResearchFormerly WizWand

Grayscale JPEG compression artifact removal on Urban100

35.77PSNRHi-IR
25.952428.501231.0533.5988Nov 27, 2024
Updated 8mo ago

Evaluation results

MethodLinks
Hi-IRQF=40, Single model trained for multiple QFs=false2024.1135.7795.61
SwinIRQF=402024.1135.7395.53
CATQF=402024.1135.7395.11
ARTQF=402024.1135.6395.66
Hi-IRQF=40, Single model trained for multiple QFs=true2024.1135.595.15
DRUNetQF=40, Single model trained for multiple QFs=true2024.1135.295.47
Hi-IRQF=30, Single model trained for multiple QFs=false2024.1134.8694.59
SwinIRQF=302024.1134.8194.42
CATQF=302024.1134.8194.49
ARTQF=302024.1134.6494.78
Hi-IRQF=30, Single model trained for multiple QFs=true2024.1134.5894.17
DRUNetQF=30, Single model trained for multiple QFs=true2024.1134.2394.14
Hi-IRQF=20, Single model trained for multiple QFs=false2024.1133.893.24
CATQF=202024.1133.5192.5
DnCNN3QF=40, Single model trained for multiple QFs=true2024.1133.4994.12
SwinIRQF=202024.1133.3892.69
ARTQF=202024.1133.2192.56
Hi-IRQF=20, Single model trained for multiple QFs=true2024.1133.1291.9
DRUNetQF=20, Single model trained for multiple QFs=true2024.1132.8192.41
DnCNN3QF=30, Single model trained for multiple QFs=true2024.1132.4793.12
Hi-IRQF=10, Single model trained for multiple QFs=false2024.1131.0789.5
JPEGQF=402024.1131.0692.15
DnCNN3QF=20, Single model trained for multiple QFs=true2024.1131.0190.5
SwinIRQF=102024.1130.8788.94
CATQF=102024.1130.8188.66
ARTQF=102024.1130.6288.08
Hi-IRQF=10, Single model trained for multiple QFs=true2024.1130.5588.35
DRUNetQF=10, Single model trained for multiple QFs=true2024.1130.3187.45
JPEGQF=302024.113090.13
JPEGQF=202024.1128.5785.45
DnCNN3QF=10, Single model trained for multiple QFs=true2024.1128.5484.84
JPEGQF=102024.1126.3378.16