Define granularity of segmented areas

Hello,

I am wondering if it is possible to determine the granularity of the segmented objects?

It seems to detect objects that are of course well defined, but very small. These small objects are not of interest of me and I would like to blend those smaller segments into larger segments.

For example, instead of segmenting a chair into 8 segments, I want to segment it as one single object.

Start by reviewing the repository's inference code and example notebooks to understand how segmented objects are produced. Determine whether the requested granularity can be controlled from the existing workflow; the issue does not specify an entry point, implementation approach, or acceptance criteria.

Written by the indexing model from the issue text.