List your company
ResearchFormerly WizWand

Image Denoising on CBSD68 sigma=25 (test)

31.72PSNRVLU-Net
23.597625.706327.81529.9237Sep 10, 2023Jan 25, 2024Jun 10, 2024Oct 25, 2024Mar 11, 2025Jul 26, 2025Dec 11, 2025
Updated 8mo ago

Evaluation results

MethodLinks
VLU-Netprotocol=one-by-one IR2025.0331.72—
PromptIRprotocol=one-by-one IR2025.0331.71—
IDRprotocol=one-by-one IR2025.0331.6—
IDRVenue=CVPR'23, Task Setting=N+H+R+B+L2025.1231.60.887
InstructIRType=All-in-one (end-to-end), Params.=16M, Setting=NHR settings2025.0331.52—
InstructIRprotocol=one-by-one IR2025.0331.52—
Restormerprotocol=one-by-one IR2025.0331.49—
VLU-NetType=deep unfolding, Params.=35M, Setting=NHR settings2025.0331.48—
AirNetprotocol=one-by-one IR2025.0331.48—
PromptIRVenue=NeurIPS'23, Task Setting=N+H+R+B+L2025.1231.470.886
GridformerVenue=IJCV'24, Task Setting=N+H+R+B+L2025.1231.450.885
Perceive-IRVenue=TIP'25, Task Setting=N+H+R+B+L2025.1231.440.887
InstructIRVenue=ECCV'24, Task Setting=N+H+R+B+L2025.1231.40.887
GridformerType=All-in-one (end-to-end), Params.=34M, Setting=NHR settings2025.0331.37—
NDRType=All-in-one (end-to-end), Params.=39M, Setting=NHR settings2025.0331.36—
AdaIRVenue=ICLR'25, Task Setting=N+H+R+B+L2025.1231.350.889
ClusIRVenue=Ours, Task Setting=N+H+R+B+L2025.1231.350.894
MoCE-IRVenue=CVPR'25, Task Setting=N+H+R+B+L2025.1231.340.887
IDRType=All-in-one (end-to-end), Params.=15M, Setting=NHR settings2025.0331.32—
DBD4Training Schema=Fixed variance (sigma=25), Training Paradigm=Noisy/Clean2023.0931.310.8548
PromptIRType=All-in-one (end-to-end), Params.=33M, Setting=NHR settings2025.0331.31—
DFPIRVenue=CVPR'25, Task Setting=N+H+R+B+L2025.1231.290.889
AirNetType=All-in-one (end-to-end), Params.=9M, Setting=NHR settings2025.0331.26—
DnCNNprotocol=one-by-one IR2025.0331.24—
FFDNetprotocol=one-by-one IR2025.0331.21—
DGUNetprotocol=one-by-one IR2025.0331.1—
N2STraining Schema=Fixed variance (sigma=25), Training Paradigm=Noisy/Noisy2023.0930.990.8405
MambaIRType=One-by-one (end-to-end), Params.=27M, Setting=NHR settings2025.0330.95—
MeDTraining Schema=Random variance (sigma in [5, 50]), Training Paradigm=Invariant Feature2023.0930.910.8573
AirNetVenue=CVPR'22, Task Setting=N+H+R+B+L2025.1230.910.882
N2CTraining Schema=Random variance (sigma in [5, 50]), Training Paradigm=Noisy/Clean2023.0930.870.8538
N2CTraining Schema=Fixed variance (sigma=25), Training Paradigm=Noisy/Clean2023.0930.830.8494
N2NTraining Schema=Random variance (sigma in [5, 50]), Training Paradigm=N2N2023.0930.770.8423
MPRNetType=One-by-one (end-to-end), Params.=16M, Setting=NHR settings2025.0330.76—
N2NTraining Schema=Fixed variance (sigma=25), Training Paradigm=N2N2023.0930.680.8334
RestormerType=One-by-one (end-to-end), Params.=26M, Setting=NHR settings2025.0330.67—
DBD4Training Schema=Random variance (sigma in [5, 50]), Training Paradigm=Noisy/Clean2023.0930.640.8491
SwinIRprotocol=one-by-one IR2025.0330.59—
MeDTraining Schema=Fixed variance (sigma=25), Training Paradigm=Invariant Feature2023.0930.570.8496
R2RTraining Schema=Fixed variance (sigma=25), Training Paradigm=Noisy/Noisy2023.0930.450.8183
TAPEVenue=ECCV'22, Task Setting=N+H+R+B+L2025.1230.180.855
LIRTraining Schema=Fixed variance (sigma=25), Training Paradigm=Invariant Feature2023.0930.010.8024
N2STraining Schema=Random variance (sigma in [5, 50]), Training Paradigm=Noisy/Noisy2023.0929.650.7801
TransWeatherVenue=CVPR'22, Task Setting=N+H+R+B+L2025.12290.841
LIRTraining Schema=Random variance (sigma in [5, 50]), Training Paradigm=Invariant Feature2023.0928.920.8069
R2RTraining Schema=Random variance (sigma in [5, 50]), Training Paradigm=Noisy/Noisy2023.0923.910.4552