I build out a keypoint skeleton to use when labelling my keypoint data for tennis court.
However I wanted to know do the edges matter? Ie does the model actually use the edges when training or at inference time?
I see other people labelling their tennis courts and some have many edges in their skeleton while others have a few.
Wanted to know the best practices for making a skeleton as well as how the edges effect the model quality and output.
FYI - I’ve already read all the docs and blog posts about skeletons.
Thanks!