Create app.py

app.py

@@ -0,0 +1,39 @@
+from diffusers import DiffusionPipeline
+import gradio as gr
+import numpy as np
+import imageio
+from PIL import Image
+import torch
+
+device = "cuda" if torch.cuda.is_available() else "cpu"
+
+pipe = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-2-inpainting")
+pipe.to(device)
+
+source_img = gr.Image(source="upload", type="numpy", tool="sketch", elem_id="source_container");
+
+def resize(height,img):
+  baseheight = height
+  img = Image.open(img)
+  hpercent = (baseheight/float(img.size[1]))
+  wsize = int((float(img.size[0])*float(hpercent)))
+  img = img.resize((wsize,baseheight), Image.LANCZOS)
+  return img
+
+def predict(source_img, prompt):
+    imageio.imwrite("data.png", source_img["image"])
+    imageio.imwrite("data_mask.png", source_img["mask"])
+    
+    src = resize(512, "data.png")
+    src.save("src.png")
+    mask = resize(512, "data_mask.png")  
+    mask.save("mask.png")
+    
+    image = pipe(prompt, image=src, mask_image=mask, strength=0.75, num_inference_steps=10).images[0]   
+    
+    return image
+
+
+title="Stable Diffusion 2.0 Inpainting CPU"
+description="Inpainting with Stable Diffusion 2.0 <br />Warning: Slow process... ~5-10 min inference time.<br> <b>Please use 512*512 or 768x768 square .png image as input to avoid memory error!!!</b>"
+gr.Interface(fn=predict, inputs=[source_img, "text"], outputs='image', title=title, description=description).launch(debug=True)