What shape to make "labels" for recognition using OpenCV?

Question

C++ / C#

Greetings!
In general, the task was to recognize marks on a sheet of paper from the camera,
I would like to know if anyone had experience, what form of marks are better recognized?
the marks themselves should be no more than 6 mm
. I saw both a circle in a circle and a square in a square and just a crosshair
. it will be easier to find.
Well, suddenly there is already an implementation of such recognition - I will be glad!

Answer 1

In general, I tried a square in a square, like a Qr code, it is
recognized remarkably!
Algorithm for OpenCV - we are looking for contours, we are looking for closed contours, we find where the contour is in the contour (you can also check that this is a square), we calculate the area of ​​\u200b\u200bthe outer and inner ones and look at the ratio.