Update app.py

app.py

@@ -6,25 +6,63 @@ import tensorflow as tf
 from transformers import SegformerForSemanticSegmentation, AutoFeatureExtractor
 import cv2
 import json
+import os
 
 # Load models
 part_seg_model = SegformerForSemanticSegmentation.from_pretrained("Mohaddz/huggingCars")
 damage_seg_model = SegformerForSemanticSegmentation.from_pretrained("Mohaddz/DamageSeg")
 feature_extractor = AutoFeatureExtractor.from_pretrained("Mohaddz/huggingCars")
 
-# Recreate the model architecture
-def create_model(input_shape, num_classes):
-    inputs = tf.keras.Input(shape=input_shape)
-    x = tf.keras.layers.Dense(64, activation='relu')(inputs)
-    x = tf.keras.layers.Dense(32, activation='relu')(x)
-    outputs = tf.keras.layers.Dense(num_classes, activation='sigmoid')(x)
-    return tf.keras.Model(inputs=inputs, outputs=outputs)
+# Attempt to load the model
+def load_model(model_path):
+    print(f"Attempting to load model from: {model_path}")
+    print(f"Current working directory: {os.getcwd()}")
+    print(f"Files in current directory: {os.listdir('.')}")
+    
+    try:
+        # Attempt 1: Load the entire model
+        model = tf.keras.models.load_model(model_path)
+        print("Successfully loaded the entire model.")
+        return model
+    except Exception as e:
+        print(f"Failed to load entire model. Error: {str(e)}")
+        
+        try:
+            # Attempt 2: Load model architecture from JSON and weights separately
+            with open(model_path.replace('.h5', '.json'), 'r') as json_file:
+                model_json = json_file.read()
+            model = tf.keras.models.model_from_json(model_json)
+            model.load_weights(model_path)
+            print("Successfully loaded model from JSON and weights.")
+            return model
+        except Exception as e:
+            print(f"Failed to load model from JSON and weights. Error: {str(e)}")
+            
+            try:
+                # Attempt 3: Load only the weights into a predefined architecture
+                input_shape = 33  # Adjust if necessary
+                num_classes = 29  # Adjust if necessary
+                inputs = tf.keras.Input(shape=(input_shape,))
+                x = tf.keras.layers.Dense(256, activation='relu')(inputs)
+                x = tf.keras.layers.Dense(128, activation='relu')(x)
+                x = tf.keras.layers.Dense(64, activation='relu')(x)
+                outputs = tf.keras.layers.Dense(num_classes, activation='sigmoid')(x)
+                model = tf.keras.Model(inputs=inputs, outputs=outputs)
+                model.load_weights(model_path)
+                print("Successfully loaded weights into predefined architecture.")
+                return model
+            except Exception as e:
+                print(f"Failed to load weights into predefined architecture. Error: {str(e)}")
+                raise Exception("All attempts to load the model failed.")
 
-# Load model weights
-input_shape = 33  # Adjust this based on your actual input shape
-num_classes = 29  # Adjust this based on your actual number of classes
-dl_model = create_model(input_shape, num_classes)
-dl_model.load_weights('improved_car_damage_prediction_model.h5')
+# Try to load the model
+try:
+    dl_model = load_model('improved_car_damage_prediction_model.h5')
+    print("Model loaded successfully.")
+    dl_model.summary()
+except Exception as e:
+    print(f"Failed to load the model: {str(e)}")
+    dl_model = None
 
 # Load parts list
 with open('cars117.json', 'r', encoding='utf-8') as f:
@@ -67,14 +105,18 @@ def process_image(image):
     input_vector = np.concatenate([part_features.mean(axis=(1, 2)), damage_features.mean(axis=(1, 2))])
     
     # Predict parts to replace using the loaded model
-    prediction = dl_model.predict(np.array([input_vector]))
-    predicted_parts = [(all_parts[i], float(prob)) for i, prob in enumerate(prediction[0]) if prob > 0.1]
-    predicted_parts.sort(key=lambda x: x[1], reverse=True)
+    if dl_model is not None:
+        prediction = dl_model.predict(np.array([input_vector]))
+        predicted_parts = [(all_parts[i], float(prob)) for i, prob in enumerate(prediction[0]) if prob > 0.1]
+        predicted_parts.sort(key=lambda x: x[1], reverse=True)
+        prediction_text = "\n".join([f"{part}: {prob:.2f}" for part, prob in predicted_parts[:5]])
+    else:
+        prediction_text = "Model failed to load. Unable to make predictions."
     
     return (Image.fromarray(annotated_image), 
             Image.fromarray(damage_heatmap_resized), 
             Image.fromarray(part_heatmap_resized), 
-            "\n".join([f"{part}: {prob:.2f}" for part, prob in predicted_parts[:5]]))
+            prediction_text)
 
 def create_heatmap(features):
     heatmap = np.sum(features, axis=0)