Update app.py

app.py

@@ -6,23 +6,20 @@ import numpy as np
 from tensorflow.keras.preprocessing.image import img_to_array, load_img
 
 # Function to load the model with custom objects
-def load_model_with_hub():
-    # Load the KerasLayer from TensorFlow Hub
-    keras_layer = hub.KerasLayer("https://tfhub.dev/google/imagenet/resnet_v2_50/classification/5", trainable=False)
-    
-    # Define the input layer
-    input_layer = tf.keras.layers.Input(shape=(224, 224, 3))
-    
-    # Pass the input through the KerasLayer
-    x = keras_layer(input_layer)
-    
-    # Build the model
-    model = tf.keras.models.Model(inputs=input_layer, outputs=x)
-    
+def load_model_with_hub(model_path):
+    # Load the model architecture without weights
+    model = tf.keras.models.load_model(model_path, compile=False)
+
+    # Define the KerasLayer from TensorFlow Hub
+    keras_layer = hub.KerasLayer("https://tfhub.dev/google/imagenet/resnet_v2_101/feature_vector/5", trainable=False)
+
+    # Add the KerasLayer to the model
+    model.add(keras_layer)
+
     return model
 
-# Loading model with custom objects
-model = load_model_with_hub(model_cat_dog.h5)
+# Loading saved model with custom objects
+model = load_model_with_hub('model_cat_dog.h5')
 
 def predict(input_image):
     try:
@@ -63,10 +60,7 @@ iface = gr.Interface(
     fn=predict, 
     inputs=gr.inputs.Image(shape=(224, 224)), 
     outputs="text",
-    title = '🐱 x 🐶 Image Recognition - Cats vs Dogs with Resnet 101 V2 🐱 x 🐶',
-    description="""<br> This model was trained to predict whether an image contains a cat or a dog. <br> 
-    <br> You can see how this model was trained on the following <a href = "https://www.kaggle.com/lusfernandotorres/computer-vision-cats-vs-dogs-w-resnet-v2-101">Kaggle Notebook</a>.
-    <br>Upload a photo to see the how the model predicts!""",
+    title = 'Image Recognition - Cats vs Dogs',
     examples = examples
 )