Open jessevdk opened 11 years ago
Hi jesse,
We do want to implement full body tracking. We gave priority to the upper body because that's arguably the most important part when using skeleton tracking and also because it was technically easier/quicker to accomplish but we definitely want to track many more joints. However, detecting the remaining joints requires a fairly good deal of investigation that at the moment I have no time to do.
Please feel free to contribute and send us patches, it would be great!
Do you have any pointers on where to start? The heuristic that you currently use comes directly from the paper or from somewhere else?
Hey Jesse,
I have some ideas but I still need to think more about whether they are feasible or not. Also, it is very difficult to explain them by writing and without any schematics. The current implementation it's based on the paper mentioned in Skeltrack's announcement post in my post. I would say it follows perhaps half of the paper tops.
If you want to hack in Skeltrack perhaps a good way to start is to watch this talk I gave at LinuxTag this year: http://www.youtube.com/watch?v=dk8EaNksAXA and of course, go through the code and see how it works.
Cheers,
Hey Joaquimrocha!
Do you happen to find the solution to full body tracking yet?
Thanks Jeet Kanjani University of Oxford
Hey, I was wondering of there were any plans to support full body skeleton tracking, i.e. lower body joints as well? Is there any technical reason why the skeleton tracking is limited to the upper body? I'm willing to contribute if it's a matter of implementation.