The video turned out to be pretty long-winded, the first few mins contain the jist of it, also in the text below :)
I am faced with a situation frequently where I have a Kinect V2 installation that is designed to have one single player interact with it, in an environment where multiple people can be in the range of Kinect. It is important in this case that the pre-processing of the Kinect Skeleton data is predictable and consistent.
Imagine that you set up interaction area bounds 1m from the Kinect, but before anyone enters that area there are 6 people lingering in the background (each being tracked) and one goes up to the designated area. You will need some processing how to select that player!
This component aims to help with that by setting up interaction bounds (configurable min/max coordinates relative to the Kinect) and only allowing one person in that area to interact.
If the person that was previously the active player leaves the area, while there are also other players in the bounds, this component will seamlessly switch to the new selected player.
This switching also works if multiple people enter the area, this component will queue them based on the order they entered.