In the Python code sample at Feature Based Image Alignment using OpenCV (C++/Python) the code does
orb = cv2.ORB_create(MAX_FEATURES) and then, later,
keypoints1, descriptors1 = orb.detectAndCompute(im1Gray, None).
This implies that
detectAndCompute is a method within the ORB class (whose constructor is
ORB_create) but OpenCV: cv::ORB Class Reference makes no mention of it.
Maybe it’s inherited from some sort of parent class but idk.
I’m just wondering because, altho using ORB works well, I found a few examples where it didn’t work so well and I was wondering if AKAZE might be better.