Analyze() fails to detect face about deepface HOT 4 CLOSED

Thanks @serengil and @AndreaLanfranchi!

Could I suggest that the "skip" option get a mention in the docstring? Here's what the docstring currently says for analyze():

        detector_backend (string): face detector backend. Options: 'opencv', 'retinaface',
            'mtcnn', 'ssd', 'dlib', 'mediapipe', 'yolov8' (default is opencv).

from deepface.

If you feed pre detected face to analyze, then set detector backend to skip.

Once a face is detected, it removes ears and it cannot extract face anymore from that manipulated image.

from deepface.

@buckeye17 to complete the answer please try to amend your last lines code to

if face_arr is not None:
    result = DeepFace.analyze(
        img_path=face_arr, 
        detector_backend = "skip",
        actions=["gender", "race"],
        )

There are basically two changes:

1. detector_backend = "skip" this says the function the image you're feeding is already a detected face hence the stack won't try to re-detect the face once again
2. actions=["gender", "race"] this tells the function which attributes to analyze. Note I've removed the "dominant_" prefix as that is part of the response. I mean when you analyze, say, "gender", you'll be returned the probability of gender man or woman and on behalf of the heaviest weight the "dominant_" attribute is also labelled.

Hope this helps.

from deepface.

Closed with PR - #1177

from deepface.