Update app.py

app.py

@@ -96,26 +96,36 @@ if opt == 'Upload image from device':
     if file:
         image = preprocess_image(file)
 
-# initialize the session state with a default value
-if 'training_mode' not in st.session_state:
-  st.session_state.training_mode = False
+# Sidebar section
+st.sidebar.title("Options")
 
-# create a button widget
-button = st.button('Toggle training mode')
+# Create a radio button widget for training mode
+training_mode = st.sidebar.radio("Select training mode", ["None", "Dropout", "Batch Normalization"])
 
-# update the session state when the button is clicked
-if button:
-  st.session_state.training_mode = not st.session_state.training_mode
+# Create a select box widget for user choice
+user_choice = st.sidebar.selectbox("Select what you want to do", ["Predict", "Train"])
 
-# display the current training mode
-st.write(f'Training mode: {st.session_state.training_mode}')
+# Display the current training mode and user choice
+st.write(f"Training mode: {training_mode}")
+st.write(f"User choice: {user_choice}")
 
 try:
     if image is not None:
         st.image(image, width=256, caption='Uploaded Image')
-        if st.button('Predict'):
-            # call the model with the training mode argument
-            prediction = model.predict(image[np.newaxis, ...], training=st.session_state.training_mode)
+        
+        # Execute different code blocks based on user choice
+        if user_choice == "Predict":
+            # Call the model with the training mode argument
+            # Use a dictionary to map the training mode to the corresponding boolean value
+            prediction = model.predict_on_batch(image[np.newaxis, ...], training={"None": False, "Dropout": True, "Batch Normalization": True}[training_mode])
             st.success(f'Prediction: {labels[np.argmax(prediction[0], axis=-1)]}')
+        elif user_choice == "Train":
+            # Generate some dummy target data for demonstration
+            # You can replace this with your actual target data
+            target = np.random.randint(0, 6, size=(1,))
+            # Call the model with the training mode argument
+            # Use a dictionary to map the training mode to the corresponding boolean value
+            loss = model.train_on_batch(image[np.newaxis, ...], target, training={"None": False, "Dropout": True, "Batch Normalization": True}[training_mode])
+            st.success(f'Loss: {loss}')
 except Exception as e:
     st.error(f"An error occurred: {e}.  Please contact us EcoClim Solutions at EcoClimSolutions.wordpress.com.")