Update app.py

app.py

@@ -22,10 +22,10 @@ st.markdown(
 This demo runs a simple **Automatic License Plate Recognition (ALPR)** pipeline 
 using models hosted on **DeGirum Cloud**.
 
-**Steps:**
-1. Upload an image with a vehicle / license plate.  
+**How to use this demo:**
+1. Upload an image containing a vehicle / license plate.  
 2. Click **Run Inference**.  
-3. See the image with detected plates and OCR labels.
+3. The result image will show detected plates with OCR labels.
 """
 )
 
@@ -33,8 +33,9 @@ using models hosted on **DeGirum Cloud**.
 # Configuration
 # -----------------------------
 hw_location = "@cloud"
-model_zoo_url = "https://cs.degirum.com/degirum/degirum"
 
+# Model zoo and model names
+model_zoo_url = "https://cs.degirum.com/degirum/degirum"
 lp_det_model_name = "yolov8n_relu6_global_lp_det--640x640_quant_n2x_orca1_1"
 lp_ocr_model_name = "yolov8s_relu6_lp_ocr_7ch--256x128_quant_n2x_orca1_1"
 
@@ -69,7 +70,7 @@ def load_compound_model():
 crop_model = load_compound_model()
 
 # -----------------------------
-# UI: upload + run
+# Upload + form
 # -----------------------------
 st.subheader("Upload an image")
 
@@ -78,28 +79,33 @@ uploaded_file = st.file_uploader(
     type=["jpg", "jpeg", "png"],
 )
 
+# Show original image as soon as it’s uploaded (simple preview)
 if uploaded_file is not None:
-    # Show original image as soon as it's uploaded
-    image = Image.open(uploaded_file).convert("RGB")
-    preview = image.copy()
+    original_image = Image.open(uploaded_file).convert("RGB")
+    preview = original_image.copy()
     preview.thumbnail((800, 800), Image.Resampling.LANCZOS)
     st.image(preview, caption="Original image", use_container_width=True)
 
-    run_button = st.button("Run Inference")
+# Use a form for the button + inference (like your original code)
+with st.form("lp_form"):
+    submitted = st.form_submit_button(
+        "Run Inference", disabled=uploaded_file is None
+    )
 
-    if run_button:
-        with st.spinner("Running license plate detection and recognition..."):
-            # Use a copy resized for the model
-            model_input = image.copy()
-            model_input.thumbnail((640, 640), Image.Resampling.LANCZOS)
+    if submitted and uploaded_file is not None:
+        # Resize image for model input
+        image_for_model = original_image.copy()
+        image_for_model.thumbnail((640, 640), Image.Resampling.LANCZOS)
 
-            results = crop_model(model_input)
+        # Run compound model
+        inference_results = crop_model(image_for_model)
 
         st.subheader("Detection result")
         st.image(
-            results.image_overlay,
+            inference_results.image_overlay,
             caption="Image with license plate bounding boxes and labels",
             use_container_width=True,
         )
-else:
+
+if uploaded_file is None:
     st.info("👈 Upload an image to get started.")