face detection and face verification inputs
Hi there, I just have a simple question, I like to know that the models inputs are in RGB or BGR??
RGB in all of our models.
https://github.com/deepinsight/insightface/search?q=ins_get_image Does it mean all these callings of ins_get_image(image) are incorrect? As to_rgb is False by default.
RGB in all of our models.
Does it mean that we have to convert image from BGR to RGB and feed to FaceAnalysis.get function?
https://github.com/deepinsight/insightface/blob/30295de48907e04077d6d22a9a8f580b525822ce/python-package/insightface/app/face_analysis.py#L58-L77
I want to know the input channel order for this specific function.
From experimentation, it seems to me that FaceAnalysis.get() function probably accepts BGR format because I can see the sex and age prediction on this image correctly when I feed BGR format to the function. If I feed RGB format to the function, it predicts female for every face in this specific image.
Please confirm that the get() function does indeed accept BGR format as input. Thanks.
I am going to necro this because its still unclear, in the examples it is shown using cv2.imread() which by default loads an image as BGR and then in the example that BGR image is fed to the model.
So is the example wrong?
If so I can open a PR to update it and also I can add a docblock to the function FacialAnalysis.get to explain what image type should be put in