Detect ArUco markers for potential use in automatic club identification

[benhalpern]

Is your feature request related to a problem? Please describe.

I personally find manually telling sim software what club I'm using to be tedious. I'm not aware of anything that does this automatically, but I believe it is very achievable with this particular system.

In general I think the focus of this project should be on efficiently achieving parity with industry-standard monitors vs feature chasing, however I think automated club detection would be such a big upgrade for a lot of use cases.

Describe the solution you'd like

ArUco markers are like minimal QR codes which could store enough data in their matrix to easily develop a club-selection algorithm and/or multiple users using a single machine. There is enough data potential for many use cases while being much more likely to be accurate than something with a more complicated process.

I envision being able to print out tiny stickers that could fit on most club heads for automatic club detection (either at address or during swing).

The white/black contrast should show up great against the strobe.

ArUco markers are already very strandard as a complement to OpenCV and I think the implementation would be straightforward.

For now, I think the resulting data detected should just be piped out via ActiveMQ to be injested for any purpose, but next steps could also include club mapping within the monitor's web app — however I think that involves some additional UX considerations which don't need to be rushed.

Describe alternatives you've considered

I'm all ears for alternative solutions to the same problem. If tiny QR codes or other barcode alternatives could be printed small enough to fit on the toe of a blade and be recognizable, more fidelity couldn't hurt. However, I suspect ArUco is going to be the most reliable and buildable solution.

Machine learning could be used to detect a club without the need of a marker for certain basic use cases, but it could never be reliable for detection of similar clubs (or even the same club bent 1 degree difference in a club fitting scenario)

Additional context

I'll reiterate that the ice-cream sundae of this project is the focused ongoing iteration towards accurate and reliable measurements with low-cost hardware, but the cherry on top when it so happens to fit so well with the existing technology.

I'm curious if anyone has thought about anything like this before.