Hey!
I’m using RealityCapture to scan a scene with some april tags. In the future application, I need to compute the pose of a webcam by again finding the april tags and using OpenCV with the known positions.
My problem is now that with my april-tag library, I get different IDs of the markers than with RealityCapture. I create some 36h11-markers with IDs from 1 to 10, but read completely different values. If found a somehow official Apriltag-repo here: GitHub - AprilRobotics/apriltag-imgs: Pre-generated AprilTag images and the images there do not fit to the tags created by RC.
Any ideas how this can happen? Is there a documentation which april-tag library is used within RC?
Hi NikolasMunich,
we have our own implementation, so differences are possible. They seem to be using a different numbering scheme. Ours is quite natural, we order the payloads in ascending order, so the first is 0x0110bba61 and the last is 0xfece0e599 (if you were to print all 587 of them).
Thanks for the fast answer!
So you mean you deviate from the coding scheme that was published here: AprilTag ?
Technically yes, it is different.
Ok. I think the april tags are a really nice feature, but to be honest, it’s not a great design to use the ‘April-tag’ name, and even the 36h11-naming scheme and then use a new code. That’s just provoking confusion and a mixup in the codes. Some people already have printed april tags and I think coding should be kept consistent with what was published under the name.
Yes, I understand to that.
I will create a feature request for this for our developers to implement/change the target creation tool.
I am sorry for the inconveniences.
1 Like
That would be great! The markers are really useful, but with a consistent naming scheme, it would be easier to use.