Semantic segmentation with HDRP(high definition render pipeline)

Good afternoon,
We am writing to you with regard to your new “high definition render pipeline” system. We hope to get some answers to our questions.
Firstly,could you tell us how we can get semantic segmentation of the picture from “high definition render pipeline” ?
Secondly, where from and in what way can we pick up these maps ?

Thank you very much for your time.

I’m searching for the same, have you got any solution yet.