[Question] How to get semantic 3d scene graph & How to run Hydra in real experiments?

[psi0000]

Thank you for yours
I hope to use this, but I have some questions.

1. I want to get 3d semantic scene graph

I found some output (later.csv .. ) , but it didn't contain the semantic graph.

example (objects_layer.csv)

nodes_active,nodes_removed,nodes_merged,nodes_w_parents,nodes_w_children,edges
0,0,0,0,0,0
1,0,0,1,0,0
1,0,0,1,0,0
1,0,0,1,0,0
1,0,0,1,0,0
1,0,0,1,0,0
1,0,0,1,0,0
1,0,0,1,0,0
1,0,0,1,0,0
1,0,0,1,0,0
1,0,0,1,0,0
1,0,0,1,0,0
1,0,0,1,0,0
1,0,0,1,0,0

I wanna change your 3d scene graph to 3d scene semantic graph.
Just like

{
  "rooms": [
    {
      "id": 1,
      "name": "Living Room",
      "objects": [
        {"id": 101, "type": "sofa", "position": [2.1, 3.5, 0]},
        {"id": 102, "type": "table", "position": [2.0, 3.0, 0]}
      ]
    },
    {
      "id": 2,
      "name": "Kitchen",
      "objects": [
        {"id": 201, "type": "refrigerator", "position": [5.0, 1.5, 0]}
      ]
    }
  ]

I hope to find the information of 3d scene graph ( I am curious whether the discovered objects' poses and the connected graphs can be semantically linked)
Somebody anything help?

2. I want to use Hydra in a real-world environment, but I can't find any testing procedures or necessary sources for this. Could you provide guidance on how to implement and test Hydra outside of simulation?