The text annotations for new scenes

[MYM181]

Hi, thanks for your great work!
I'm curious about the text descriptions of each object for training/testing.
Is it a manual or automatic annotation process? When training a new scene, I think it's not feasible to manually write descriptions about each object first.
How to ensure that the text annotations of each object contain enough knowledge and various descriptions about it?
Hope to get your reply soon. Thanks!

[heshuting555]

Thanks for the interest!
For now, the object descriptions are written manually. Automating or semi-automating this step is definitely on our roadmap.

[MYM181]

Thanks for the interest! For now, the object descriptions are written manually. Automating or semi-automating this step is definitely on our roadmap.

Sure. Thanks!