hhhhnwl/PointRend-simple-pytorch

The missing parts about this line

Opened this issue · 1 comments

out,mask = net(up_fine, up_coarse)

Looks like up_fine and up_coarse is fine_grained and coarse_prediction.
I also have simple question for real usage about this. Do we have to train this code separately after we train our network?
Why the network has two return?

@freaad did you find your answer?