Update app.py

app.py

@@ -90,6 +90,7 @@ model = load_model()
     
 
 # Utility Functions
+
 def upload_to_gcs(image_data: io.BytesIO, filename: str, content_type='application/dicom'):
     """Uploads an image to Google Cloud Storage."""
     try:
@@ -245,8 +246,17 @@ def grad_cam(input_model, img_array, cls, layer_name):
 
 # Compute Grad-CAM
 def compute_gradcam(model, img_path, layer_name='bn'):
+    base_model  = keras.applications.DenseNet121(weights = './densenet.hdf5', include_top = False)
+    x = base_model.output
+    x = GlobalAveragePooling2D()(x)
+    predictions = Dense(14, activation = "sigmoid")(x)
+    model_gradcam = Model(inputs=base_model.input, outputs=predictions)
+    model_gradcam.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=0.0001),
+              loss='sparse_categorical_crossentropy')
+    model.load_weights('./pretrained_model.h5')
+    
     preprocessed_input = load_image(img_path)
-    predictions = model.predict(preprocessed_input)
+    predictions = model_gradcam.predict(preprocessed_input)
 
     original_image = load_image(img_path, preprocess=False)
 
@@ -342,10 +352,6 @@ def redirect_button(url):
     if button:
         st.markdown(f'<meta http-equiv="refresh" content="0;url={url}" />', unsafe_allow_html=True)
 
-def load_model():
-    model = tf.keras.models.load_model('./model_renamed.h5',custom_objects={'DepthwiseConv2D': tf.keras.layers.DepthwiseConv2D})
-    return model
-
 ###########################################################################################
 ########################### Streamlit Interface ###########################################
 ###########################################################################################