chore(deps): update machine-learning (#6302)

* chore(deps): update machine-learning

* fix typing, use new lifespan syntax

* wrap in try / finally

* move log

---------

Co-authored-by: renovate[bot] <29139614+renovate[bot]@users.noreply.github.com>
Co-authored-by: mertalev <101130780+mertalev@users.noreply.github.com>

machine-learning/app/models/facial_recognition.py

@@ -5,9 +5,10 @@ import cv2
 import numpy as np
 from insightface.model_zoo import ArcFaceONNX, RetinaFace
 from insightface.utils.face_align import norm_crop
+from numpy.typing import NDArray
 
 from app.config import clean_name
-from app.schemas import BoundingBox, Face, ModelType, ndarray_f32
+from app.schemas import Face, ModelType, is_ndarray
 
 from .base import InferenceModel
 
@@ -36,22 +37,25 @@ class FaceRecognizer(InferenceModel):
         )
         self.rec_model.prepare(ctx_id=0)
 
-    def _predict(self, image: ndarray_f32 | bytes) -> list[Face]:
+    def _predict(self, image: NDArray[np.uint8] | bytes) -> list[Face]:
         if isinstance(image, bytes):
-            image = cv2.imdecode(np.frombuffer(image, np.uint8), cv2.IMREAD_COLOR)
-        bboxes, kpss = self.det_model.detect(image)
+            decoded_image = cv2.imdecode(np.frombuffer(image, np.uint8), cv2.IMREAD_COLOR)
+        else:
+            decoded_image = image
+        assert is_ndarray(decoded_image, np.uint8)
+        bboxes, kpss = self.det_model.detect(decoded_image)
         if bboxes.size == 0:
             return []
-        assert isinstance(image, np.ndarray) and isinstance(kpss, np.ndarray)
+        assert is_ndarray(kpss, np.float32)
 
         scores = bboxes[:, 4].tolist()
         bboxes = bboxes[:, :4].round().tolist()
 
         results = []
-        height, width, _ = image.shape
+        height, width, _ = decoded_image.shape
         for (x1, y1, x2, y2), score, kps in zip(bboxes, scores, kpss):
-            cropped_img = norm_crop(image, kps)
-            embedding: ndarray_f32 = self.rec_model.get_feat(cropped_img)[0]
+            cropped_img = norm_crop(decoded_image, kps)
+            embedding: NDArray[np.float32] = self.rec_model.get_feat(cropped_img)[0]
             face: Face = {
                 "imageWidth": width,
                 "imageHeight": height,