Generating high-res images using ESRGAN-PyTorch

This repository show the code to upscale pictures by 4x using ESRGAN.

Given a low resolution picture, the goal is to upscale by 4x and compare with the Bicubic Algorithm.

As you can see, the results are very good. Try yourself!

Demo

Click here to view in 1080p

Endpoint available

Endpoint	Description
http://localhost:8000/	Front-end to perform style transfer.

Install

1. Clone this repository

git clone https://github.com/renatoviolin/super-resolution-image-esrgan.git
cd super-resolution-image-esrgan

2. Download the pre-trained model

wget "https://www.dropbox.com/s/58x64ex9m047gdy/model.pth?dl=0" -O esrgan/model.pth

3. Install dependencies

pip install -r requirements.txt

4. Start web-application

cd web-app
uvicorn app:app --host 0.0.0.0 --port 8000

Running on Google Colab

You can running on colab using ngrok.

References

ESRGAN (Enhanced SRGAN): https://github.com/xinntao/ESRGAN

BibTeX

@InProceedings{wang2018esrgan,
    author = {Wang, Xintao and Yu, Ke and Wu, Shixiang and Gu, Jinjin and Liu, Yihao and Dong, Chao and Qiao, Yu and Loy, Chen Change},
    title = {ESRGAN: Enhanced super-resolution generative adversarial networks},
    booktitle = {The European Conference on Computer Vision Workshops (ECCVW)},
    month = {September},
    year = {2018}
}