search

xianlin7 / SAMUS

MIT License
143 stars 13 forks source link

dice and loss #10

Open yyy1998-i opened 7 months ago

yyy1998-i commented 7 months ago

May I ask why I started with a high validation set and why the test DICE was very low? 74176832af46645a4b0899ebe8fc098b

yyy1998-i commented 7 months ago

I'm stupid, bother, there are a few classes, id is the number of classes, not the background, for example, a tumor is 1, id=1

xianlin7 commented 5 months ago

I'm stupid, bother, there are a few classes, id is the number of classes, not the background, for example, a tumor is 1, id=1

Yes, ID is the label value of each category in the corresponding dataset. We have provided a more detailed explanation of ID in SAMCT