How to use the dataset

I downloaded the dataset and found there is no mask annotation. Should I use the `bbox` or `point_coords` to produce ground truth mask?

Start by reading the dataset documentation and the example notebooks described in the repository, then trace how bbox and point_coords are used. Done means documenting whether either input can serve as ground-truth mask annotation and explaining the correct source or workflow for obtaining masks.

Written by the indexing model from the issue text.