Vision Processing in Command Based Java

Hey all,
This year we are doing vision processing for the first time in java. We are already having problems getting the stream to display to the smart dashboard. We are using a command based template. I copied the code from the sample camera code into the init() of the robot.java class, but it says something about how a network tables instance is already instantiated. We know the camera is functioning because we deployed the sample vision code and that worked perfectly.

Furthermore, if we are going to go deeper into vision processsing, where do we put the vision code? Should it be put in a separate subsystem or in a separate project entirely like I’ve seen some teams do? I am having trouble finding examples relevant to our code template.How the structure change if we were to offload processing to a raspberry pi? We are using GRIP for processing. Thanks for the help in advance.