Replies: 1 comment
-
Hi, @caumente. You can use a sampler and the queue. Please take a look at the docs for Patch-based pipelines and let us know if that would help. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hey everyone,
First of all, thanks for developing TorchIO.
Currently I am working on a project related to multi-modal medical imaging segmentation on 3D. I load the images and their segmentation and so that I get two numpy.ndarrays with shapes:
image = ndarray(4, 155, 240, 240)
mask = ndarray(155, 240, 240)
In order to do data augmentation I would like to crop randomly the image and the mask to (128, 128, 128) spatial dimensions. Does cropOrpad do it randomly or centering? I suppose that if the crop is random, I couldn't crop them as follow
I also wonder if it would be better do this kind of transformation into the dataloader or while training step on fly..
Thanks!
Beta Was this translation helpful? Give feedback.
All reactions