In the path "confil/rtmdet/", all rtmdet instance segmentation config files rely on the rtmdet_l_8xb32-300e_coco python file, which use '../base/datasets/coco_detection.py' as base config file.
But in the other model config files, such as mask rcnn, there is the '../base/datasets/coco_instance.py' instead.
So why using rtmdet to do segmentation task need the coco detection config?
In the path "confil/rtmdet/", all rtmdet instance segmentation config files rely on the rtmdet_l_8xb32-300e_coco python file, which use '../base/datasets/coco_detection.py' as base config file.
But in the other model config files, such as mask rcnn, there is the '../base/datasets/coco_instance.py' instead.
So why using rtmdet to do segmentation task need the coco detection config?