mattmacy/vnet.pytorch

Can dice loss be used for multi-class or rewriten for multi-class?

elias-1 opened this issue · 1 comments

Can dice loss be used for multi-class or rewriten for multi-class?

Yes, it is easy to manipulate it to work for mutli-class problem. Check http://www.niftynet.io/. Could you use dice loss on multi-Gpu?