Hi,
I'm currently computing SIFT descriptors in Python on custum keypoints using the following code :
cv2descriptor = cv2.DescriptorExtractor_create("SIFT")
kp, des = cv2descriptor.compute(img,custumkeypoints)
I'm wondering how the SIFT are computed. The original SIFT algorithm uses a 16x16 pixels neighbourhood around the center of the keypoint and split it into 4x4 area.
Do the SIFT computation on custom keypoints which diameter can be variable works the same way, splitting the area in 4*4 squares?
Thanks