feat: implement FaceEngine service with Haar Cascade detection, eye-based alignment, and ArcFace embedding extraction

app/services/face_recognition.py

@@ -114,6 +114,19 @@ class FaceEngine:
         face_img = cv2.resize(face_img, (112, 112))
         return face_img
 
+    def _normalize_brightness(self, image_bgr: np.ndarray) -> np.ndarray:
+        """Apply CLAHE to normalize lighting sensitivity."""
+        try:
+            lab = cv2.cvtColor(image_bgr, cv2.COLOR_BGR2LAB)
+            l, a, b = cv2.split(lab)
+            clahe = cv2.createCLAHE(clipLimit=3.0, tileGridSize=(8, 8))
+            cl = clahe.apply(l)
+            limg = cv2.merge((cl, a, b))
+            return cv2.cvtColor(limg, cv2.COLOR_LAB2BGR)
+        except Exception as e:
+            logger.warning(f"Brightness normalization failed: {e}")
+            return image_bgr
+
     def _preprocess_face(self, image_bgr: np.ndarray, x: int, y: int, w: int, h: int, eye_centers: Optional[List[np.ndarray]] = None) -> np.ndarray:
         """Preprocess face for ArcFace: Align, Crop, Resize, Normalize."""
         if eye_centers and len(eye_centers) >= 2:
@@ -143,6 +156,9 @@ class FaceEngine:
             "box": None
         }
 
+        # Apply brightness normalization for better low-light handling
+        image_bgr = self._normalize_brightness(image_bgr)
+
         
         try:
             if self.detector is None: