search

SuperMedIntel / MedSegDiff

Medical Image Segmentation with Diffusion Model
MIT License
1.09k stars 166 forks source link

BRATS model generating only black Samples #155

Open qahsinahk opened 9 months ago

qahsinahk commented 9 months ago

I trained the model on BRATS2020 for about 220000 iterations, now, when I run the sample file, it generates only black samples. It also generates samples, but they are not really visible I have no idea what could be the reason behind it, can anybody help me with it? I will be very grateful. Moreover, how can I calculate the performance of the model trained on BRATS.

Thanks!

@WuJunde

VLAD-KONATA commented 9 months ago

i have the same problem, the sample performed badly

HopinNotDisaster commented 9 months ago

O!My God!

VLAD-KONATA commented 9 months ago

So what did you do for that? @VLAD-KONATA

i have no idea to explain or solve this problem, so i also need the author's help.😩

WuJunde commented 9 months ago

have you ensembled the predictions to get the final results?

Taocunguo commented 8 months ago

我想我没有这样做，因为我不知道。如果你能告诉我，我将不胜感激。 谢谢！ Is there a solution to this problem? I have also encountered this problem

Ding-yd commented 6 months ago

Like you, I did the same work and didn't perform well on the Brats dataset

ErfanNourian commented 3 weeks ago

@WuJunde the problem is that each sample is nearly black and ensembling them won't help much.