-
Notifications
You must be signed in to change notification settings - Fork 1.1k
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
Port deepmedic #94
Comments
This might warrant a separate issue but it's probably worth highlignting that the deepmedic paper relies of a post-processing by a dense CRF. The dense CRF is implemented through a permutohedral lattice approach. Although deemedic itself does not rely on backpropagating through the CRF, other papers do (e.g. CRF-RNN) and they also rely on the same permutohedral lattice approach. As such, it would be good to have a generic permutohedral lattice / CRF-RNN module in MONAI. A pytorch implementation of the CRF-RNN can be found here but note that the permutohedral lattice does not run on GPU there: The corresponding NiftyNet module is here but lacks some GPU ops and cannot back-propagate through features: An alternative implementation of a GPU-based backpropagatable permutohedral lattice can be found here: This might also be interesting: |
Hello, |
Hi @brosscle currently there's no plan from the core dev team to port this architectures, please feel free to submit a PR. for your (or anyone interested in contributing) reference, there are a few examples about using the relevant monai APIs:
also there's a generic contributing guide: https://github.com/Project-MONAI/MONAI/blob/master/CONTRIBUTING.md |
I will take a look at it, thanks ! |
closed as merged @charliebudd |
a subtask of #78, port deepmedic
ref:
https://niftynet.readthedocs.io/en/dev/niftynet.network.deepmedic.html
The text was updated successfully, but these errors were encountered: