Create app.py

app.py

@@ -0,0 +1,37 @@
+import os
+os.system("pip install gfpgan")
+import gradio as gr
+from PIL import Image
+import torch
+import cv2
+import glob
+import numpy as np
+from basicsr.utils import imwrite
+from gfpgan import GFPGANer
+
+bg_upsampler = None
+
+# set up GFPGAN restorer
+restorer = GFPGANer(
+    model_path='GFPGANCleanv1-NoCE-C2.pth',
+    upscale=2,
+    arch='clean',
+    channel_multiplier=2,
+    bg_upsampler=bg_upsampler)
+
+
+def inference(img):
+    input_img = cv2.imread(img, cv2.IMREAD_COLOR)
+    cropped_faces, restored_faces, restored_img = restorer.enhance(
+        input_img, has_aligned=False, only_center_face=False, paste_back=True)
+
+    return Image.fromarray(restored_faces[0][:, :, ::-1])
+
+
+title = "GFP-GAN FACE ENHANCE MODEL"
+gr.Interface(
+    inference,
+    [gr.inputs.Image(type="filepath", label="Input")],
+    gr.outputs.Image(type="pil", label="Output"),
+    title=title
+).launch(enable_queue=True)