Update app.py

app.py

@@ -32,16 +32,11 @@ RECOGNITION_THRESHOLD = 0.1 # Adjust as needed
 
 # Preprocess the image
 def preprocess_image(image):
-    # Resize image to match FaceNet input size (160x160)
-    image = cv2.resize(image, (160, 160))
-
-    # Normalize the image to a range of [-1, 1] as FaceNet might expect
-    image = image.astype('float32')
-    mean, std = image.mean(), image.std()
-    image = (image - mean) / std
-
-    # Expand dimensions to add batch size of 1
-    return np.expand_dims(image, axis=0)
+    image = cv2.resize(image, (160, 160))  # Resize image to match FaceNet input size
+    image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)  # Convert to RGB (OpenCV loads images in BGR)
+    image = tf.convert_to_tensor(image)  # Convert to TensorFlow tensor
+    image = tf.image.convert_image_dtype(image, tf.float32)  # Normalize pixel values
+    return np.expand_dims(image, axis=0)  # Add batch dimension
 
 # Generate embedding
 def generate_embedding(image):