app.py

import os
import sys
from torchvision.transforms import functional
sys.modules["torchvision.transforms.functional_tensor"] = functional

from basicsr.archs.srvgg_arch import SRVGGNetCompact
from gfpgan.utils import GFPGANer
from realesrgan.utils import RealESRGANer

import torch
import cv2
import gradio as gr

# 필수 모델 다운로드
if not os.path.exists('realesr-general-x4v3.pth'):
    os.system("wget https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.5.0/realesr-general-x4v3.pth -P .")
if not os.path.exists('GFPGANv1.2.pth'):
    os.system("wget https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.2.pth -P .")
if not os.path.exists('GFPGANv1.3.pth'):
    os.system("wget https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.3.pth -P .")
if not os.path.exists('GFPGANv1.4.pth'):
    os.system("wget https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.4.pth -P .")
if not os.path.exists('RestoreFormer.pth'):
    os.system("wget https://github.com/TencentARC/GFPGAN/releases/download/v1.3.4/RestoreFormer.pth -P .")

model = SRVGGNetCompact(num_in_ch=3, num_out_ch=3, num_feat=64, num_conv=32, upscale=4, act_type='prelu')
model_path = 'realesr-general-x4v3.pth'
half = True if torch.cuda.is_available() else False
upsampler = RealESRGANer(scale=4, model_path=model_path, model=model, tile=0, tile_pad=10, pre_pad=0, half=half)

# 이미지 저장 디렉토리 생성 (필요시 주석 해제)
# os.makedirs('output', exist_ok=True)

def upscaler(img, version, scale):
    try:
        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:
            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)

        face_enhancer = GFPGANer(
            model_path=f'{version}.pth',
            upscale=2,
            arch='RestoreFormer' if version=='RestoreFormer' else 'clean',
            channel_multiplier=2,
            bg_upsampler=upsampler
        )

        try:
            _, _, output = face_enhancer.enhance(img, has_aligned=False, only_center_face=False, paste_back=True)
        except RuntimeError as error:
            print('오류', 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('잘못된 재스케일링 입력.', error)

        output = cv2.cvtColor(output, cv2.COLOR_BGR2RGB)
        return output
    except Exception as error:
        print('전역 예외', error)
        return None, None

if __name__ == "__main__":
    title = "이미지 업스케일 및 복원 [GFPGAN 알고리즘]"

    demo = gr.Interface(
            upscaler, [
                gr.Image(type="filepath", label="입력"),
                gr.Radio(['GFPGANv1.2', 'GFPGANv1.3', 'GFPGANv1.4', 'RestoreFormer'], type="value", label="버전", value="GFPGANv1.4", visible=False),
                gr.Number(label="재스케일링 계수", value=0, visible=False),
            ], [
                gr.Image(type="numpy", label="출력"),
            ],
            title=title,
            examples=[["예제.png", "GFPGANv1.4", 0]],
            allow_flagging="never"
        )

    demo.queue()
    demo.launch()