Update src/streamlit_app.py

src/streamlit_app.py

@@ -13,7 +13,6 @@ from tensorflow.keras.models import Sequential
 from tensorflow.keras.layers import Conv2D, MaxPooling2D, Flatten, Dense
 
 tf.config.set_visible_devices([], 'GPU')
-st.write("TensorFlow version:", tf.__version__)
 
 
 # ---------------------------
@@ -170,6 +169,8 @@ def fetch_image():
 
 # ---------------------------
 
+placeholder = st.empty()
+
  
 
 st.title("Neural Network Classification Demo")
@@ -199,100 +200,100 @@ if 'model' not in st.session_state:
     st.session_state.calculating=False
 
 
+with placeholder.container():
 
-
-# Button to fetch a new image
-col1, col2, col3 = st.columns([1,1,1])
-
-if st.session_state.current_image is not None and st.session_state.started is False:
-    st.session_state.started=True
-
-with col1: 
-    if st.button("Start",disabled=st.session_state.started):
-        if not st.session_state.started:
-            fetch_image()
-
-
-# Display the current image
-
-if st.session_state.current_image is not None:
-    with col2:
-        if st.button("cat"):#,disabled=st.session_state.calculating):
-            print("cat pressed")
-            st.session_state.label_input="cat"
-    with col3:
-        if st.button("dog"):#,disabled=st.session_state.calculating):
-            print("dog pressed")
-            st.session_state.label_input="dog"
-    print(f"SHAPE:{st.session_state.current_image.shape}")
-    prediction = st.session_state.model.predict(np.array([st.session_state.current_image]))[0][0]
-
-    st.session_state.current_prediction = 'dog' if prediction > 0.5 else 'cat'
-    st.success(f"**Model Predicts:** {st.session_state.current_prediction} --- (cat-confidence {(1-prediction)*100:.2f}%; dog-confidence {(prediction)*100:.2f}%)")
-
-    st.image(st.session_state.unprocessed_image)
+    # Button to fetch a new image
+    col1, col2, col3 = st.columns([1,1,1])
     
-
- 
-
-    # User input for label
- 
-
-    if st.session_state.label_input in ['cat', 'dog']:
-        label_input=st.session_state.label_input
-        st.session_state.label_input="None"
-
-        # Convert user input to 0 (cat) or 1 (dog)
-        print(f"LABEL CLICKED IS: {label_input.lower()}")
-        label = 0 if label_input.lower() == 'cat' else 1
-
-        st.session_state.current_label = label
-
- 
-
-        # Add the labeled image and label to training data
-
-        st.session_state.training_data.append((st.session_state.current_image, label))
-
- 
-
-        # Retrain the model
-
-        image = np.array([img for img, _ in st.session_state.training_data])
-
-        label = np.array([lab for _, lab in st.session_state.training_data])
-
+    if st.session_state.current_image is not None and st.session_state.started is False:
+        st.session_state.started=True
+    
+    with col1: 
+        if st.button("Start",disabled=st.session_state.started):
+            if not st.session_state.started:
+                fetch_image()
+    
+    
+    # Display the current image
+    
+    if st.session_state.current_image is not None:
+        with col2:
+            if st.button("cat"):#,disabled=st.session_state.calculating):
+                print("cat pressed")
+                st.session_state.label_input="cat"
+        with col3:
+            if st.button("dog"):#,disabled=st.session_state.calculating):
+                print("dog pressed")
+                st.session_state.label_input="dog"
+        print(f"SHAPE:{st.session_state.current_image.shape}")
+        prediction = st.session_state.model.predict(np.array([st.session_state.current_image]))[0][0]
+    
+        st.session_state.current_prediction = 'dog' if prediction > 0.5 else 'cat'
+        st.success(f"**Model Predicts:** {st.session_state.current_prediction} --- (cat-confidence {(1-prediction)*100:.2f}%; dog-confidence {(prediction)*100:.2f}%)")
+    
+        st.image(st.session_state.unprocessed_image)
         
-
- 
-
-        # Predict the current image
-
-        st.session_state.current_image=None
-
-        print("before model fit")
-        def model_fit():
-            print("Entering model fit function")
-            st.session_state.model.fit(image, label, epochs=1)
-            st.write(st.session_state.model.evaluate(image,  label, verbose=2))
-            st.session_state.calculating=True
-            st.write("hi")
-            return
-        model_fit()
-
-if st.session_state.calculating:
     
-    print("after model fit")
-    #st.session_state.unprocessed_image=None
-    print("before fetch_image")
-
-    fetch_image()
-    print("after fetch_image")
-    st.info(f"You clicked on last picture (picture {st.session_state.which_pic}): {label_input}")
-    st.session_state.which_pic=st.session_state.which_pic+1
-    st.session_state.calculating=False
-
+     
+    
+        # User input for label
+     
+    
+        if st.session_state.label_input in ['cat', 'dog']:
+            label_input=st.session_state.label_input
+            st.session_state.label_input="None"
+    
+            # Convert user input to 0 (cat) or 1 (dog)
+            print(f"LABEL CLICKED IS: {label_input.lower()}")
+            label = 0 if label_input.lower() == 'cat' else 1
+    
+            st.session_state.current_label = label
+    
+     
+    
+            # Add the labeled image and label to training data
+    
+            st.session_state.training_data.append((st.session_state.current_image, label))
+    
+     
+    
+            # Retrain the model
+    
+            image = np.array([img for img, _ in st.session_state.training_data])
+    
+            label = np.array([lab for _, lab in st.session_state.training_data])
+    
+            
+    
+     
+    
+            # Predict the current image
+    
+            st.session_state.current_image=None
+    
+            print("before model fit")
+            def model_fit():
+                print("Entering model fit function")
+                st.session_state.model.fit(image, label, epochs=1)
+                st.write(st.session_state.model.evaluate(image,  label, verbose=2))
+                st.session_state.calculating=True
+                st.write("hi")
+                return
+            model_fit()
+    
+    if st.session_state.calculating:
         
+        print("after model fit")
+        #st.session_state.unprocessed_image=None
+        print("before fetch_image")
+    
+        fetch_image()
+        print("after fetch_image")
+        st.info(f"You clicked on last picture (picture {st.session_state.which_pic}): {label_input}")
+        st.session_state.which_pic=st.session_state.which_pic+1
+        st.session_state.calculating=False
+    
+