Add test.

Add comments.
Update color format for `__find_bulk_embeddings`.

deepface/modules/recognition.py

@@ -398,6 +398,7 @@ def __find_bulk_embeddings(
                 enforce_detection=enforce_detection,
                 align=align,
                 expand_percentage=expand_percentage,
+                color_face='bgr'  # `represent` expects images in bgr format.
             )
 
         except ValueError as err:

deepface/modules/representation.py

@@ -74,6 +74,7 @@ def represent(
     # we have run pre-process in verification. so, this can be skipped if it is coming from verify.
     target_size = model.input_shape
     if detector_backend != "skip":
+        # Images are returned in RGB format.
         img_objs = detection.extract_faces(
             img_path=img_path,
             detector_backend=detector_backend,
@@ -87,13 +88,13 @@ def represent(
     else:  # skip
         # Try load. If load error, will raise exception internal
         img, _ = image_utils.load_image(img_path)
-        
-        # bgr to rgb
-        img = img[:, :, ::-1]
-        
+
         if len(img.shape) != 3:
             raise ValueError(f"Input img must be 3 dimensional but it is {img.shape}")
 
+        # Convert to RGB format to keep compatability with `extract_faces`.
+        img = img[:, :, ::-1]
+
         # make dummy region and confidence to keep compatibility with `extract_faces`
         img_objs = [
             {

tests/test_represent.py

@@ -81,3 +81,23 @@ def test_max_faces():
     max_faces = 1
     results = DeepFace.represent(img_path="dataset/couple.jpg", max_faces=max_faces)
     assert len(results) == max_faces
+
+
+def test_represent_detector_backend():
+    # Results using a detection backend.
+    results_1 = DeepFace.represent(img_path="dataset/img1.jpg")
+    assert len(results_1) == 1
+
+    # Results performing face extraction first.
+    faces = DeepFace.extract_faces(img_path="dataset/img1.jpg", color_face='bgr')
+    assert len(faces) == 1
+
+    # Images sent into represent need to be in BGR format.
+    img = faces[0]['face']
+    results_2 = DeepFace.represent(img_path=img, detector_backend="skip")
+    assert len(results_2) == 1
+
+    # The embeddings should be the exact same for both cases.
+    embedding_1 = results_1[0]['embedding']
+    embedding_2 = results_2[0]['embedding']
+    assert embedding_1 == embedding_2