What does the mask parameter refer to?

[Wang-Shuo]

GHM_Detection/mmdetection/mmdet/core/loss/ghm_loss.py

Line 24 in ff58c3e

def calc(self, input, target, mask):

[libuyu]

Some of the candidate boxes are ignored for training such as crowded boxes in COCO. And the mask here means whether this candidate box is used (mask[i,j]=1) or ignored (mask[i,j]=0). It just acts as loss weights.

[Wang-Shuo]

I see. Thanks a lot.

[milliema]

May I ask where is the mask (now named as label_weight) calculated?