lazy load model

app.py

@@ -50,16 +50,15 @@ h1 {
 
 rng = np.random.default_rng(0xDEADBEEF)
 
-
-def load_ahoy() -> AHOY:
-    """Load and warmup the model."""
-    model = load_model("ahoy-RGB-b2")
-    return model
-
+model: AHOY = None
 
 @spaces.GPU
-def inference(model: AHOY, image):
+def inference(image):
     """Run inference on image and return annotated image."""
+    global model
+    if model is None:
+        # lazy load model
+        model = load_model("ahoy-RGB-b2")
     results = model(image)
     return results.draw(image, diameter=4)
 
@@ -110,7 +109,7 @@ with gr.Blocks(theme=theme, css=css, title="SEA.AI Vision Demo") as demo:
     # event listeners
     img_url.change(load_image_from_url, [img_url], img_input)
     submit.click(check_image, [img_input], None, show_api=False).success(
-        lambda image: inference(model, image),
+        inference,
         [img_input],
         img_output,
         api_name="inference",