In the original paper,
"the mask target is the intersection between an RoI and its associated ground-truth mask." .
But it seems that the code here: https://github.com/CharlesShang/FastMaskRCNN/blob/master/libs/layers/mask.py#L73-L76just use the associated ground-truth mask,not the intersection between an RoI and its associated ground-truth mask, as the "mask_target". So is there any bug in the code , or have I understand something wrong?
@CharlesShang
In the original paper, "the mask target is the intersection between an RoI and its associated ground-truth mask." . But it seems that the code here: https://github.com/CharlesShang/FastMaskRCNN/blob/master/libs/layers/mask.py#L73-L76 just use the associated ground-truth mask, not the intersection between an RoI and its associated ground-truth mask, as the "mask_target". So is there any bug in the code , or have I understand something wrong? @CharlesShang