Multi class Focal-loss

fastai
1

Hi all,
I would like to know if exists a multi class Focal-loss function/class for semantic segmentation, compatible with Fastai libraries.

I found this implementation, but I can not understand the meaning of the term V(y).