Resize and normalize input image before processing

app.py

@@ -75,20 +75,26 @@ with gr.Blocks(css=css) as demo:
                                type='numpy', elem_id='img-display-input')
         depth_image_slider = ImageSlider(
             label="Depth Map with Slider View", elem_id='img-display-output', position=0.5,)
-    raw_file = gr.File(
-        label="16-bit raw depth (can be considered as disparity)")
+    raw_file = gr.File(label="Download Depth Map")
     submit = gr.Button("Submit")
 
     def on_submit(image):
         original_image = image.copy()
 
-        # This is for resizing the image to 518x518
+        # Resize the image if it is larger than 640x480
+        max_width, max_height = 640, 480
         h, w = image.shape[:2]
+        if w > max_width or h > max_height:
+            scaling_factor = min(max_width / w, max_height / h)
+            image = cv2.resize(image, None, fx=scaling_factor,
+                               fy=scaling_factor, interpolation=cv2.INTER_AREA)
 
+        # Normalize the image
         image = np.asarray(image, dtype=np.float32) / 255.0
         image = torch.from_numpy(image.transpose((2, 0, 1)))
         image = Normalize(mean=[0.485, 0.456, 0.406], std=[
             0.229, 0.224, 0.225])(image)
+
         with torch.no_grad():
             image = torch.autograd.Variable(image.unsqueeze(0))
             print("== Processing image")
@@ -104,7 +110,6 @@ with gr.Blocks(css=css) as demo:
 
             # Continue with your file saving operations
             tmp = tempfile.NamedTemporaryFile(suffix='.png', delete=False)
-            # cv2.imwrite(tmp.name, output_image)
             plt.imsave(tmp.name, pred_depth, cmap='jet')
 
             return [(original_image, tmp.name), tmp.name]