Training with Customised Dataset

[SoumiDas]

Hello,

Is it possible to use this training model for semantically segmenting images from a custom made dataset where we have RGB Image and a ground truth data alongside with say 12 classes?

It'll be really helpful to receive responses.