search

MontaEllis / Pytorch-Medical-Segmentation

This repository is an unoffical PyTorch implementation of Medical segmentation in 2D and 3D.
MIT License
864 stars 196 forks source link

the prediction results are all black #45

Closed ARnnn closed 2 years ago

ARnnn commented 2 years ago

I fixed the errors in the model prediction in the same way as you, but now the prediction results are all black. What's the problem. What I did was a dichotomy. During the test, there was such a warning: image

ARnnn commented 2 years ago

I solved the above problem, but the predicted result looks very bad image the tensorboard is as follows： ![Uploading image.png…]()

MontaEllis commented 2 years ago

I can't see your tensorboard.

ARnnn commented 2 years ago

I can't see your tensorboard.

image

MontaEllis commented 2 years ago

Can I see your hparam.py? I guess you set 'croporpad' in a wrong way.

ARnnn commented 2 years ago

I set the same parameters of hparam.py as you

ARnnn commented 2 years ago

I haven't changed anything except my own dataset

MontaEllis commented 2 years ago

You should modify 'croporpad' at your own dataset.

ARnnn commented 2 years ago

Different cases in my data have different sizes，so how can I set the croporpad according to my own data?

ARnnn commented 2 years ago

In many slices in my case, some contain the parts to be segmented and some do not (the label is all black). Do I need to remove the slices that do not contain the parts to be segmented?

MontaEllis commented 2 years ago

Yes.