cszn/scunet 🖼️❓ → ❓

▶️ 102.0K runs 📅 May 2022 ⚙️ Cog 0.2.0 🔗 GitHub 📄 Paper ⚖️ License
image-denoising image-restoration

About

Practical Blind Denoising via Swin-Conv-UNet and Data Synthesis

Example Output

Output

{"denoised_image":"https://replicate.delivery/mgxm/d5d01de0-c13d-4f28-879e-68525879ceae/output.png"}

Performance Metrics

1.88s Prediction Time
2.13s Total Time
All Input Parameters 
{
  "image": "https://replicate.delivery/mgxm/fbbc7700-69a6-4be8-8a08-41a75037066f/Flowers.png",
  "model_name": "real image denoising"
}
Input Parameters
image (required) Type: string
Input image.
model_name Default: real image denoising
Choose a model. 15, 25 and 50 in grayscale images and color images correspond to the added noise level, and the output will show image_with_added_noise and denoised_image.
Output Schema
denoised_image Type: stringFormat: uri
Denoised Image
image_with_added_noise Type: stringFormat: uri
Image With Added Noise
Example Execution Logs 
Model params number: 17946072
Version Details
Version ID
df9a3c1dbc6c1f7f4c2d244f68dffa2699a169cf5e701e0d6a009bf6ff507f26
Version Created
May 4, 2022
Run on Replicate →