I found several examples, such as here , here and here , in which it is recommended to use a shared memory structure or use sockets.
I think that in this case, transferring your programs through sockets would be a better idea, since your applications will not be closely connected, so you just need to set the IP address, port and a set of commands.
According to this , it is possible to create a C ++ server on Kinect, but apart from this I cannot say much, since I have never worked on projects related to Kinect.
source share