Low-light Image Enhancement on SID (test)
28.88PSNRSID
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| SIDTime=0.51s, Params=7.76M2019.08 | 28.88 | 0.79 | 0.36 | |
| Ours (Enhancement-Net variant)Time=0.48s, Params=0.88M2019.08 | 27.96 | 0.77 | 0.36 | |
| PTG-RM (SNR backbone)Prior=CLIP, Backbone=SNR2024.03 | 23.34 | 0.63 | — | |
| PTG-RM (SNR backbone)Mode=Feature Refinement, Backbone=SNR2024.03 | 23.17 | 0.636 | — | |
| PTG-RM (SNR backbone)Prior=BLIP2, Backbone=SNR2024.03 | 23.15 | 0.627 | — | |
| PTG-RM (SNR backbone)Prior=Stable Diffusion, Backbone=SNR2024.03 | 23.08 | 0.631 | — | |
| PTG-RM (SNR backbone)Prior=Restoration models (SDSD), Backbone=SNR2024.03 | 23.06 | 0.632 | — | |
| VCR2026.03 | 23.012 | 0.712 | — | |
| CIDNet2026.03 | 22.904 | 0.676 | — | |
| SNRRefinement=None2024.03 | 22.87 | 0.625 | — | |
| PTG-RM (URetinex backbone)Mode=Feature Refinement, Backbone=URetinex2024.03 | 22.4 | 0.626 | — | |
| PTG-RM (URetinex backbone)Prior=CLIP, Backbone=URetinex2024.03 | 22.34 | 0.623 | — | |
| PTG-RM (UHD backbone)Prior=Stable Diffusion, Backbone=UHD2024.03 | 22.21 | 0.625 | — | |
| PTG-RM (URetinex backbone)Prior=Stable Diffusion, Backbone=URetinex2024.03 | 22.21 | 0.623 | — | |
| PTG-RM (URetinex backbone)Prior=Restoration models (SDSD), Backbone=URetinex2024.03 | 22.17 | 0.625 | — | |
| PTG-RM (URetinex backbone)Prior=BLIP2, Backbone=URetinex2024.03 | 22.02 | 0.621 | — | |
| URetinexRefinement=None2024.03 | 21.56 | 0.619 | — | |
| PTG-RM (UHD backbone)Prior=Restoration models (SDSD), Backbone=UHD2024.03 | 21.11 | 0.618 | — | |
| PTG-RM (UHD backbone)Mode=Feature Refinement, Backbone=UHD2024.03 | 21.08 | 0.619 | — | |
| PTG-RM (UHD backbone)Prior=BLIP2, Backbone=UHD2024.03 | 21.06 | 0.619 | — | |
| PTG-RM (UHD backbone)Prior=CLIP, Backbone=UHD2024.03 | 20.99 | 0.616 | — | |
| UHDRefinement=None2024.03 | 20.46 | 0.614 | — | |
| SCI + Ours-cRefinement Module=Ours-c, Pre-trained Model=CLIP2024.03 | 19.61 | 0.598 | — | |
| SCIRefinement Module=None2024.03 | 19.09 | 0.585 | — | |
| RUAS + Ours-cRefinement Module=Ours-c, Pre-trained Model=CLIP2024.03 | 18.93 | 0.59 | — | |
| ZeroDCE + Ours-cRefinement Module=Ours-c, Pre-trained Model=CLIP2024.03 | 18.65 | 0.593 | — | |
| RUASRefinement Module=None2024.03 | 18.44 | 0.581 | — | |
| ZeroDCERefinement Module=None2024.03 | 18.08 | 0.576 | — | |
| LLFlow2026.03 | 16.226 | 0.367 | — | |
| RetinexNet2026.03 | 15.695 | 0.395 | — | |
| ZeroDCE2026.03 | 14.087 | 0.09 | — | |
| RUAS2026.03 | 12.622 | 0.081 | — |