Update app.py

app.py

@@ -1,8 +1,10 @@
 import streamlit as st
-from transformers import pipeline
-import requests
-from io import BytesIO
+from tensorflow.keras.models import load_model
+from tensorflow.keras.utils import CustomObjectScope
+from tensorflow.keras.initializers import glorot_uniform
 from PIL import Image
+import numpy as np
+import io
 
 # Define models and their validation accuracies
 model_options = {
@@ -16,13 +18,35 @@ model_options = {
     }
 }
 
-# Load the model from Hugging Face repo
-def load_model(model_path):
-    # Here you would use the Hugging Face `transformers` library to load your model.
-    # However, since these are `.h5` models (likely Keras models), use the appropriate loader.
-    # This example assumes you have a custom loader function for Keras models.
-    from tensorflow.keras.models import load_model
-    return load_model(model_path)
+# Load the model with custom objects if necessary
+def load_model_with_custom_objects(model_path):
+    if not os.path.isfile(model_path):
+        raise FileNotFoundError(f"Model file not found: {model_path}")
+
+    custom_objects = {
+        'GlorotUniform': glorot_uniform
+        # Add other custom objects if needed
+    }
+
+    try:
+        with CustomObjectScope(custom_objects):
+            model = load_model(model_path)
+    except Exception as e:
+        st.error(f"Error loading model: {str(e)}")
+        raise
+
+    return model
+
+# Image preprocessing (adjust as needed for your model)
+def preprocess_image(image):
+    # Convert image to grayscale and resize
+    image = image.convert('L')  # Convert to grayscale if necessary
+    image = image.resize((64, 64))  # Resize to match the model input shape
+    image_array = np.array(image)
+    image_array = image_array.astype('float32') / 255.0  # Normalize
+    image_array = np.expand_dims(image_array, axis=0)  # Add batch dimension
+    image_array = np.expand_dims(image_array, axis=-1)  # Add channel dimension if needed
+    return image_array
 
 def main():
     st.title("Pneumonia Detection App")
@@ -32,8 +56,12 @@ def main():
     model_accuracy = model_options[model_name]["accuracy"]
 
     # Load the selected model
-    model = load_model(model_path)
-    
+    try:
+        model = load_model_with_custom_objects(model_path)
+    except Exception as e:
+        st.error(f"Failed to load model: {e}")
+        return
+
     st.write(f"Model: {model_name}")
     st.write(f"Validation Accuracy: {model_accuracy}%")
 
@@ -44,15 +72,15 @@ def main():
         st.image(image, caption="Uploaded Image", use_column_width=True)
 
         # Perform prediction using the model
-        # This part depends on how your model expects input.
-        # Here, you would preprocess the image and perform prediction.
-        # For example:
-        # img_array = preprocess_image(image)
-        # prediction = model.predict(img_array)
-        # st.write("Prediction:", prediction)
-
-        # Example placeholder for prediction output
-        st.write("Prediction: [Placeholder for actual prediction]")
+        img_array = preprocess_image(image)
+        
+        try:
+            prediction = model.predict(img_array)
+            predicted_class = "Pneumonia" if prediction[0][0] > 0.5 else "Normal"
+            st.write(f"Prediction: {predicted_class}")
+        except Exception as e:
+            st.error(f"Error during prediction: {str(e)}")
 
 if __name__ == "__main__":
     main()
+