donnyyou/torchcv

SFNET: what's difference between x,fpn_dsn?

jinmingyi1998 opened this issue · 1 comments

x, fpn_dsn = self.head(x_)

而且,ResSFNet 和 AlignHead 都有self.fpn_out,是否是重复的?(虽然这两个fpn_out有一点点差别)

论文中有两次FAM(upsample和cat) ,而代码中只有上采样时做了?