Motion Deblurring on FFHQ and ImageNet 256x256 resolution (val)
29.17PSNROurs(AI)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Ours(AI)Inference Paradigm=Amortized inference, Diffusion Model=Unconditional Diffusion Model2026.05 | 29.17 | 85 | 0.004 | |
| RED-DiffInference Paradigm=Particle variational inference, Diffusion Model=Unconditional Diffusion Model2026.05 | 29.02 | 84 | 0.001 | |
| Ours(VI)Inference Paradigm=Particle variational inference, Diffusion Model=Unconditional Diffusion Model2026.05 | 28.18 | 83 | 0.009 | |
| DAVIInference Paradigm=Amortized inference, Diffusion Model=Unconditional Diffusion Model2026.05 | 27.69 | 82 | 0.003 | |
| ΠGDMInference Paradigm=Sampling, Diffusion Model=Unconditional Diffusion Model2026.05 | 25.82 | 75 | 0.006 | |
| RLSDInference Paradigm=Particle variational inference, Diffusion Model=Stable Diffusion2026.05 | 25.64 | 82 | 0.002 | |
| DPSInference Paradigm=Sampling, Diffusion Model=Unconditional Diffusion Model2026.05 | 20.34 | 55 | 0.006 |