Update app.py

app.py

@@ -66,11 +66,23 @@ def resize_image(input_image, resolution):
     img = cv2.resize(input_image, (W, H), interpolation=cv2.INTER_LANCZOS4 if k > 1 else cv2.INTER_AREA)
     return img
 
+# once user upload an image, the original image is stored in `original_image`
+def store_img(img):
+    # image upload is too slow
+    if min(img.shape[0], img.shape[1]) > 1024:
+        img = resize_image(img, 1024)
+    if max(img.shape[0], img.shape[1])*1.0/min(img.shape[0], img.shape[1])>2.0:
+        raise gr.Error('image aspect ratio cannot be larger than 2.0')
+    return img
+
 def process(original_image, input_mask, prompt, negative_prompt, blended, invert_mask, control_strength, seed, randomize_seed, guidance_scale, num_inference_steps):
     if original_image is None:
         raise gr.Error('Please upload the input image')
     if input_mask is None:
         raise gr.Error("Please upload a white-black Mask image")
+    #resizing input image and mask of the object
+    original_image = store_img(original_image)
+    input_mask = store_img(input_mask)
     
     H, W = original_image.shape[:2]
     original_mask = cv2.resize(input_mask, (W, H))
@@ -129,21 +141,6 @@ with gr.Blocks() as demo:
         
         with gr.Column():
             result_gallery = gr.Gallery(label='Output', show_label=False, elem_id="gallery", preview=True)
-
-    # once user upload an image, the original image is stored in `original_image`
-    def store_img(img):
-        # image upload is too slow
-        if min(img.shape[0], img.shape[1]) > 1024:
-            img = resize_image(img, 1024)
-        if max(img.shape[0], img.shape[1])*1.0/min(img.shape[0], img.shape[1])>2.0:
-            raise gr.Error('image aspect ratio cannot be larger than 2.0')
-        return img, None  # when new image is uploaded, `selected_points` should be empty
-
-    original_image.upload(
-        store_img,
-        [original_image],
-        [original_image]
-    )
     
     inputs = [original_image, input_mask, prompt, negative_prompt, blended, invert_mask, control_strength, seed, randomize_seed, guidance_scale, num_inference_steps]
     run_button.click(fn=process, inputs=inputs, outputs=[result_gallery])