app.py

from fastapi import FastAPI
import os
import shutil
import uuid
import cv2
import gradio as gr
import torch
from basicsr.archs.rrdbnet_arch import RRDBNet
from gfpgan.utils import GFPGANer
from realesrgan.utils import RealESRGANer
import uvicorn

# download weights for RealESRGAN
if not os.path.exists('model_zoo/real/RealESRGAN_x4plus.pth'):
    os.system("wget https://github.com/xinntao/Real-ESRGAN/releases/download/v0.1.0/RealESRGAN_x4plus.pth -P model_zoo/real")
if not os.path.exists('model_zoo/gan/GFPGANv1.4.pth'):
    os.system("wget https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.4.pth -P model_zoo/gan")
if not os.path.exists('model_zoo/swinir/003_realSR_BSRGAN_DFO_s64w8_SwinIR-M_x4_GAN.pth'):
    os.system('wget https://github.com/JingyunLiang/SwinIR/releases/download/v0.0/003_realSR_BSRGAN_DFO_s64w8_SwinIR-M_x4_GAN.pth -P model_zoo/swinir')

model = RRDBNet(num_in_ch=3, num_out_ch=3, num_feat=64, num_block=23, num_grow_ch=32, scale=4)
model_path = 'model_zoo/real/RealESRGAN_x4plus.pth'
netscale = 4
tile = 400 if torch.cuda.is_available() else 0
dni_weight = None
# restorer
upsampler = RealESRGANer(
    scale=netscale,
    model_path=model_path,
    dni_weight=dni_weight,
    model=model,
    tile=tile,
    tile_pad=10,
    pre_pad=0,
    half=False, #Use fp32 precision during inference. Default: fp16 (half precision).
    gpu_id=None) #gpu device to use (default=None) can be 0,1,2 for multi-gpu

def inference(img, scale):
    # background enhancer with RealESRGAN
    os.makedirs('output', exist_ok=True)
    if scale > 4:
        scale = 4  # avoid too large scale value
    try:
        extension = os.path.splitext(os.path.basename(str(img)))[1]
        img = cv2.imread(img, cv2.IMREAD_UNCHANGED)
        if len(img.shape) == 3 and img.shape[2] == 4:
            img_mode = 'RGBA'
        elif len(img.shape) == 2:  # for gray inputs
            img_mode = None
            img = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
        else:
            img_mode = None

        h, w = img.shape[0:2]
        if h < 300:
            img = cv2.resize(img, (w * 2, h * 2), interpolation=cv2.INTER_LANCZOS4)
        try:
            face_enhancer = GFPGANer(
                model_path='model_zoo/gan/GFPGANv1.4.pth', upscale=scale, arch='clean', channel_multiplier=2, bg_upsampler=upsampler)
            _, _, output = face_enhancer.enhance(img, has_aligned=False, only_center_face=False, paste_back=True)


        except RuntimeError as error:
            print('Error', error)

        try:
            if scale != 2:
                interpolation = cv2.INTER_AREA if scale < 2 else cv2.INTER_LANCZOS4
                h, w = img.shape[0:2]
                output = cv2.resize(output, (int(w * scale / 2), int(h * scale / 2)), interpolation=interpolation)
        except Exception as error:
            print('wrong scale input.', error)
        if img_mode == 'RGBA':  # RGBA images should be saved in png format
            extension = 'png'
        else:
            extension = 'jpg'

        filename = str(uuid.uuid4())    
        save_path = f'output/out_{filename}.{extension}'
        cv2.imwrite(save_path, output)

        output = cv2.cvtColor(output, cv2.COLOR_BGR2RGB)
        return output, save_path
    except Exception as error:
        print('global exception', error)
        return None, None
title = "Real Esrgan Restore Ai Face Restoration by appsgenz.com"
description = ""
article = "AppsGenz"
grApp = gr.Interface(
    inference, [
        gr.Image(type="filepath", label="Input"),
        gr.Number(label="Rescaling factor. Note max rescaling factor is 4", value=2),
    ], [
        gr.Image(type="numpy", label="Output (The whole image)"),
        gr.File(label="Download the output image")
    ],
    title=title,
    description=description,
    article=article)
grApp.queue(concurrency_count=2)
grApp.launch()