Use Blind Super Resolution model DCLS-SR to super-resolve image degradation

I want to clean a degraded image

[Image: input→output]

From this small image, it has been scaled up four times to the size above.

Blind Super Resolution is a technology for dealing with various image degradations.

Using the State of Art model

Paper with Code’s Blind Super-Resolution on Set5–4x upscaling Let’s use the top DCLS-SR model in the field.

GitHub - megvii-research/DCLS-SR: Official PyTorch implementation of the paper "Deep Constrained…

procedure

Clone the repository.

Get Pre-trained Weight from the repository page on GitHub.

Install Pytorch==1.6 (Other versions didn’t work well with Fourier transform functions).

Specify the setting file of the model (version and how many times) you want to use with -opt, and rewrite the Pre-trained Weight Path in the setting file.

result

[Image: input→output]

🐣

I’m a freelance engineer.
Work consultation
Please feel free to contact us with a brief development description.
rockyshikoku@gmail.com

I am making an app that uses Core ML and ARKit.
We send machine learning / AR related information.

GitHub

Twitter

Medium