Update app.py

app.py

@@ -5,9 +5,6 @@ import streamlit as st
 st.set_page_config(page_title="Image Classifier", layout="wide")
 
 # Now import other libraries
-os.system("pip install tensorflow")
-os.system("pip install scikit-learn")
-
 import tensorflow as tf
 import numpy as np
 import pickle
@@ -19,12 +16,17 @@ LABEL_ENCODER_PATH = "le.pkl"
 EXPECTED_SIZE = (64, 64)  # Update this based on your model's input shape
 
 def load_resources():
-    """Load model and label encoder."""
+    """Load model and label encoder with custom object handling."""
     try:
-        # Try loading with custom InputLayer configuration
+        # Define custom objects to handle compatibility issues
         custom_objects = {
-            'InputLayer': lambda **kwargs: tf.keras.layers.InputLayer(**{k: v for k, v in kwargs.items() if k != 'batch_shape'})
+            # Handle InputLayer batch_shape issue
+            'InputLayer': lambda **kwargs: tf.keras.layers.InputLayer(**{k: v for k, v in kwargs.items() if k != 'batch_shape'}),
+            # Handle DTypePolicy issue
+            'DTypePolicy': tf.keras.mixed_precision.Policy
         }
+        
+        # Try loading with custom objects
         model = tf.keras.models.load_model(
             MODEL_PATH,
             compile=False,
@@ -32,6 +34,7 @@ def load_resources():
         )
     except Exception as e:
         st.error(f"Error loading model: {str(e)}")
+        st.error("Please ensure you're using TensorFlow 2.x and the model file is not corrupted.")
         return None, None
     
     try:
@@ -92,5 +95,6 @@ if uploaded_file:
         if model is None or label_encoder is None:
             st.error("Model or label encoder failed to load. Please check the files.")
         else:
-            prediction = predict(image)
-            st.success(f"🎯 Predicted Class: {prediction}")
+            with st.spinner('Predicting...'):
+                prediction = predict(image)
+                st.success(f"🎯 Predicted Class: {prediction}")