Cover a tabletop with trackables

I'm making a table-top game, so I want to track about 1 squared metre.

I've tested using a 1 single large image with trackable squares (12 6x6 cm squares with detailed pattern. An example is attached.) spread-out over the table. I expected that seeing only one of the squares would suffice for the camera to detect the image, which is the case. However, if I tell the app to only track 1 specific square, I notice a huge tracking improvement.

Am I going about this the wrong way? Can I perhaps give the app all of the squares individually, but still have 1 model placed on them all? It seems that having to constantly look for the whole image is having a negative effect.

I have read through the best practices on image detection.

I appreciate any help.