app.py

import os
import shutil
import torch
import cv2
import gradio as gr
from PIL import Image

os.system("pip install einops")
os.system("git clone https://github.com/swz30/Restormer.git")
shutil.copy("demo_gradio.py","Restormer/demo_gradio.py")
os.chdir('Restormer')

# Download pretrained models
os.system("wget https://github.com/swz30/Restormer/releases/download/v1.0/real_denoising.pth -P Denoising/pretrained_models")
os.system("wget https://github.com/swz30/Restormer/releases/download/v1.0/single_image_defocus_deblurring.pth -P Defocus_Deblurring/pretrained_models")
os.system("wget https://github.com/swz30/Restormer/releases/download/v1.0/motion_deblurring.pth -P Motion_Deblurring/pretrained_models")
os.system("wget https://github.com/swz30/Restormer/releases/download/v1.0/deraining.pth -P Deraining/pretrained_models")

# Download sample images
os.system("wget https://github.com/swz30/Restormer/releases/download/v1.0/sample_images.zip -P demo")
shutil.unpack_archive('demo/sample_images.zip', 'demo/')
os.remove('demo/sample_images.zip')


examples = [['demo/sample_images/Real_Denoising/degraded/117355.png', 'Denoising'],
            ['demo/sample_images/Single_Image_Defocus_Deblurring/degraded/engagement.jpg', 'Defocus Deblurring'],
            ['demo/sample_images/Motion_Deblurring/degraded/GoPro-GOPR0854_11_00-000090-input.jpg','Motion Deblurring'],
            ['demo/sample_images/Deraining/degraded/Rain100H-77-input.jpg','Deraining']]


title = "Restormer"
description = """
Gradio demo for Restormer: Efficient Transformer for High-Resolution Image Restoration, CVPR 2022--ORAL. <a href='https://arxiv.org/abs/2111.09881'>[Paper]</a><a href='https://github.com/swz30/Restormer'>[Github Code]</a>\n 
With Restormer, you can perform: (1) Image Denoising, (2) Defocus Deblurring, (3)  Motion Deblurring, and (4) Image Deraining. 
To use it, simply upload your own image, or click one of the examples provided below.
"""
article = "<p style='text-align: center'><a href='https://arxiv.org/abs/2111.09881'>Restormer: Efficient Transformer for High-Resolution Image Restoration </a> | <a href='https://github.com/swz30/Restormer'>Github Repo</a></p>"


def inference(img,task):
    os.system('mkdir temp')
    max_res = 904
    width, height = img.size
    if max(width,height) > max_res:
      scale = min(width,height)/max(width,height)
      if width > max_res:
        width = max_res
        height = int(scale*max_res)
      if height > max_res:
        height = max_res
        width = int(scale*max_res)
      img = img.resize((width,height), Image.ANTIALIAS)
  
    img.save("temp/image.jpg", "JPEG")

    if task == 'Motion Deblurring':
      task = 'Motion_Deblurring'
      os.system("python demo_gradio.py --task 'Motion_Deblurring' --input_path 'temp/image.jpg' --result_dir './temp/'")
  
    if task == 'Defocus Deblurring':
      task = 'Single_Image_Defocus_Deblurring'
      os.system("python demo_gradio.py --task 'Single_Image_Defocus_Deblurring' --input_path 'temp/image.jpg' --result_dir './temp/'")
  
    if task == 'Denoising':
      task = 'Real_Denoising'
      os.system("python demo_gradio.py --task 'Real_Denoising' --input_path 'temp/image.jpg' --result_dir './temp/'")
  
    if task == 'Deraining':
      os.system("python demo_gradio.py --task 'Deraining' --input_path 'temp/image.jpg' --result_dir './temp/'")
  
    return f'temp/{task}/image.jpg'
    
gr.Interface(
    inference,
    [
        gr.inputs.Image(type="pil", label="Input"),
        gr.inputs.Radio(["Denoising", "Defocus Deblurring", "Motion Deblurring", "Deraining"], default="Denoising", label='task type')
    ],
    gr.outputs.Image(type="file", label="Output"),
    title=title,
    description=description,
    article=article,
    examples=examples,
    allow_flagging=False,
    ).launch(debug=False,enable_queue=True)