Downloads · 30 days
0
dlp-nppe3-lowlight-denoise
harshsroniyar/dlp-nppe3-lowlight-denoise
dlp-nppe3-lowlight-denoise is a machine learning model from harshsroniyar. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
RRDB-backbone, PSNR-oriented (Charbonnier + small MS-SSIM aux loss) model for joint denoising and 4x super-resolution of low-light images.
Access
Public
Updated Sep 6, 2026
Repo size
279 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.pth288 MB · 87%
At a glance
- Access
- Public
- Created
- Sep 6, 2026
- Updated
- Sep 6, 2026
- SHA
- 6d65261f
Model card
From the Hugging Face model README
Low-Light Denoising + 4x Super-Resolution (RRDB + ECA)
RRDB-backbone, PSNR-oriented (Charbonnier + small MS-SSIM aux loss) model for joint denoising and 4x super-resolution of low-light images.
- Params: 11.96M
- Best validation PSNR: 39.205 dB (tiled inference + x8 self-ensemble TTA)
- Scale: 4x
Usage
import torch
ckpt = torch.load("model_ema_final.pth", map_location="cpu")
# rebuild RRDBNet with the same cfg saved in ckpt["cfg"], then:
model.load_state_dict(ckpt["state_dict"])
- Created
- Sep 6, 2026
- Updated
- Sep 6, 2026