losses_pytorch vs test/loss_functions
daandres opened this issue · 2 comments
Hi,
thanks for this great collection of segmentation losses.
But I wonder, what are the differences between losses_pytorch
(https://github.com/JunMa11/SegLoss/tree/master/test/loss_functions) and test/loss_functions
(https://github.com/JunMa11/SegLoss/tree/master/test/loss_functions)?
And which one would recommend for custom test?
I saw that test/loss_functions
have dependencies to nnU-Net, while the others don't.
Thank you
Hi @daandres,
Thanks for your interest.
- If you want to use the loss functions in your own network, I would recommend https://github.com/JunMa11/SegLoss/tree/master/losses_pytorch
- On the other hand, I tested all the loss functions with nnU-Net V1 (https://github.com/JunMa11/SegLoss/tree/master/test) .
Now, nnU-Net has moved to V2 but I have not created a new version for the loss functions. If you want to use nnU-Net V2 with different loss functions, I would recommend https://github.com/MIC-DKFZ/nnUNet/tree/master/nnunet/training/network_training/nnUNet_variants/loss_function