Update app.py

app.py

@@ -26,16 +26,7 @@ from peft import PeftModel
 from gradcam_xception import generate_smoothgrad_visualizations_xception
 warnings.filterwarnings("ignore", category=UserWarning)
 
-# Define Xception transform function directly in app.py
-def get_xception_transform():
-    """Get the image transformation pipeline for Xception input."""
-    # Standard Xception preprocessing
-    transform = transforms.Compose([
-        transforms.Resize((299, 299)),
-        transforms.ToTensor(),
-        transforms.Normalize(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5]),
-    ])
-    return transform
+# Xception transform is now defined directly in preprocess_image_xception
 
 # App title and description
 st.set_page_config(
@@ -357,18 +348,32 @@ class ImageDataset(torch.utils.data.Dataset):
             face_img = Image.fromarray(face_img_np)
             
             # Apply transform to face image
+            IMAGE_SIZE = 299
             if self.transform:
                 face_tensor = self.transform(face_img)
             else:
-                face_tensor = transforms.ToTensor()(face_img)
+                # Use default transform if none provided
+                transform = transforms.Compose([
+                    transforms.Resize((IMAGE_SIZE, IMAGE_SIZE)),
+                    transforms.ToTensor(),
+                    transforms.Normalize(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5]),
+                ])
+                face_tensor = transform(face_img)
             
             return face_tensor, label, "uploaded_image", original_image, face_box, self.dataset_name
         else:
             # Process the whole image
+            IMAGE_SIZE = 299
             if self.transform:
                 image_tensor = self.transform(self.image)
             else:
-                image_tensor = transforms.ToTensor()(self.image)
+                # Use default transform if none provided
+                transform = transforms.Compose([
+                    transforms.Resize((IMAGE_SIZE, IMAGE_SIZE)),
+                    transforms.ToTensor(),
+                    transforms.Normalize(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5]),
+                ])
+                image_tensor = transform(self.image)
                 
             return image_tensor, label, "uploaded_image", original_image, None, self.dataset_name
 
@@ -387,7 +392,12 @@ def process_image_with_xception_gradcam(image, model, device, pred_class):
         raw_cam, cam_img, overlay, comparison = cam_results
         
         # Extract the face box from the dataset if needed
-        transform = get_xception_transform()
+        IMAGE_SIZE = 299
+        transform = transforms.Compose([
+            transforms.Resize((IMAGE_SIZE, IMAGE_SIZE)),
+            transforms.ToTensor(),
+            transforms.Normalize(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5]),
+        ])
         dataset = ImageDataset(image, transform=transform, face_only=True)
         _, _, _, _, face_box, _ = dataset[0]
         
@@ -399,11 +409,12 @@ def process_image_with_xception_gradcam(image, model, device, pred_class):
 # ----- Xception Model Loading -----
 @st.cache_resource
 def load_detection_model_xception():
-    """Loads the Xception model from our module"""
+    """Loads the Xception model from HF Hub."""
     with st.spinner("Loading Xception model for deepfake detection..."):
         try:
             log_debug("Beginning Xception model loading")
             from gradcam_xception import load_xception_model
+            log_debug("Loading Xception model (this may take a moment)...")
             model = load_xception_model()
             
             # Get the device
@@ -412,7 +423,7 @@ def load_detection_model_xception():
             
             model.to(device)
             model.eval()
-            log_debug("Xception model loaded successfully")
+            log_debug(f"Xception model loaded to {device}.")
             return model, device
         except ImportError as e:
             st.error(f"Import Error: {str(e)}. Make sure gradcam_xception.py is present.")
@@ -695,9 +706,15 @@ def preprocess_image_xception(image):
             face_img_for_transform = Image.fromarray(image_np[y1:y2, x1:x2])
             # Use the original detected box (without padding) for display rectangle
             face_box_display = (x, y, w, h)
+            log_debug(f"Face detected: Box {face_box_display}")
 
         # Xception specific transform
-        transform = get_xception_transform()
+        IMAGE_SIZE = 299
+        transform = transforms.Compose([
+            transforms.Resize((IMAGE_SIZE, IMAGE_SIZE)),
+            transforms.ToTensor(),
+            transforms.Normalize(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5]), # Standard Xception norm
+        ])
         # Apply transform to the selected region (face or whole image)
         input_tensor = transform(face_img_for_transform).unsqueeze(0)