Update app.py

app.py

@@ -5,11 +5,15 @@ import tensorflow as tf
 import tensorflow_hub as hub
 import cv2
 
-# Ensure compatibility between TensorFlow and TensorFlow Hub
-tf.keras.utils.get_custom_objects()['KerasLayer'] = hub.KerasLayer
+# Register KerasLayer from TensorFlow Hub
+tf.keras.utils.get_custom_objects().update({'KerasLayer': hub.KerasLayer})
 
-# Load the model with the correct path and custom objects
-model = tf.keras.models.load_model('bird_model4.h5', custom_objects={'KerasLayer': hub.KerasLayer})
+# Load the model and specify custom objects
+try:
+    model = tf.keras.models.load_model('bird_model4.h5', custom_objects={'KerasLayer': hub.KerasLayer})
+except ValueError as e:
+    print("Error loading the model:", e)
+    exit()
 
 # Load class labels from the text file
 train_info = []
@@ -18,20 +22,20 @@ with open('labelwithspace.txt', 'r') as file:
 
 # Function to preprocess the image and make predictions
 def predict_image(image):
-    # Resize the image to the expected input shape
+    # Resize and normalize the image
     img = cv2.resize(image, (224, 224))
-    img = img / 255.0  # Normalize the image to the range [0, 1]
+    img = img / 255.0
     
-    # Make predictions using the loaded model
+    # Make predictions
     predictions = model.predict(img[np.newaxis, ...])[0]
-    top_classes = np.argsort(predictions)[-3:][::-1]  # Get indices of top 3 predictions
-    top_class = top_classes[0]  # Index of the highest prediction
-    label = train_info[top_class]  # Retrieve the class label using the index
+    top_classes = np.argsort(predictions)[-3:][::-1]  # Indices of top 3 predictions
+    top_class = top_classes[0]  # Index of the highest probability class
+    label = train_info[top_class]  # Get the corresponding label
     return label
 
 # Define the Gradio interface
 input_image = gr.inputs.Image(shape=(224, 224))
 output_label = gr.outputs.Label()
 
-# Launch the Gradio app
+# Launch the Gradio interface
 gr.Interface(fn=predict_image, inputs=input_image, outputs=output_label, capture_session=True).launch()