-
Notifications
You must be signed in to change notification settings - Fork 1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Cannot run V-Net on medical decathlon data #7852
Comments
Hi @linnabraham, looks like a shape mismatch issue. Did you try to check your input data shape before sending to the model? |
@KumoLiu I did now and it seems like the decathlon data shape is not compatible with the V-Net. I was not expecting that since I had earlier used the same data with tensorflow implementation of V-Net (https://github.com/NVIDIA/DeepLearningExamples/tree/master/TensorFlow/Segmentation/VNet). But I found an issue with the V-Net implementation. It seems like the |
The MONAI/monai/networks/nets/vnet.py Line 211 in 64ea76d
In Tensorflow, they also set out_channels as 16: https://github.com/NVIDIA/DeepLearningExamples/blob/729963dd47e7c8bd462ad10bfac7a7b0b604e6dd/TensorFlow/Segmentation/VNet/model/vnet.py#L34 |
Thanks for pointing out the tensorflow code. But I am still confused. My input has shape (64, 128, 128). Right now I edited the source code to remove 16 from being hard coded, but no matter what I give as out_channel, 1, 16, 64, 128, I am getting a shape mismatch error. What do I do? |
If your shape is (64, 128, 128), then your |
Thanks for pointing that out. I set it to 2. I could not use 16 as out channels, so I tried 64 itself. Now I get this error
|
Describe the bug
PyTorch complains of size mismatch when using V-Net with medical decathlon data.
To Reproduce
Expected behavior
Training happens
Screenshots
Complete Traceback
Environment
Ensuring you use the relevant python executable, please paste the output of:
The text was updated successfully, but these errors were encountered: