Is it worth ot use Panoptic segmentation in this case?

[homuhumushomu]

I am trying to analyze a set of images that consists on well defined objects (screwdrivers, hammers, nails, etc) sparsely on top a table. Sometimes the floor is also visible (i.e. the boundaries of the table are within the image). I need to find out:

• the position and type of each object (within a set)
• the bounds of the table (if visible)
• the type of table (within a set)

What I have sucessfully achieved is training an Instance Segmentation model with MASK RCNN and detected each object. I was now going to extract pieces of the image in which objects are not present, match them agains the tables and through a majority vote find out which table it is.

But I was wondering if a panoptic segmentation will do the trick directly. Any hints or ideas?