Clarification on the evaluation

Hi authors (@kampta @abhijeet00 ) ,

The paper mentions that "We use the official validation split from COCO as test set in our experiments". Can you please clarify if you are referring to the COCO instance dataset or the COCO panoptic dateset?