princeton-vl/RAFT

How about use deconvolution instead of convex upsampling in paper?

ducheng678 opened this issue · 1 comments

saikrn112 commented

I am curious about this as well, did you get an answer for this?