'--ndet' flag about seglossodyssey HOT 2 CLOSED

Hi @abergsneider ,

Thanks for your interest very much.

I guess you are using the nnU-Net V2. However, my repo was developed with nnU-Net V1.

If you want to use the Dice loss to train your model, I would recommend nnUNetTrainerV2_Loss_Dice.py.

nnUNet_train 3d_lowres nnUNetTrainerV2_Loss_Dice TaskXXX_MYTASK FOLD

Hope this answer can help you. Also, please feel free to raise any questions.

Best regards,
Jun

from seglossodyssey.

Thanks @JunMa11, I will go ahead and play around with those loss functions. And thanks for kindly pointing me in the right direction.

Best,

Andres

from seglossodyssey.