You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I‘m trying to find a „deformable 3d convolutional layer“ for pytorch for 3d grayscale images (in my case medical ultrasound data).
I‘ve build an auto-encoder using https://github.com/kondratevakate/3d-deformable-convolutions in the encoder and decoder. Unfortunately it seems not to learn properly (compared to a typical network with non-deformable convolutional kernels it learns slowly).
Now I would like to try your implementation.
I couldn‘t figure out how to format the „offset“-parameter of the forward function of the deformable layer.
You‘ve mentioned in the README that currently only your core part of the code is published. Is the deformable layer currently ready-to-use?
Would be glad to hear from you!
Best regards
Moritz
The text was updated successfully, but these errors were encountered:
Hey,
I‘m trying to find a „deformable 3d convolutional layer“ for pytorch for 3d grayscale images (in my case medical ultrasound data).
I‘ve build an auto-encoder using https://github.com/kondratevakate/3d-deformable-convolutions in the encoder and decoder. Unfortunately it seems not to learn properly (compared to a typical network with non-deformable convolutional kernels it learns slowly).
Also tried this one:
https://github.com/XinyiYing/D3Dnet
Similar issue there.
Now I would like to try your implementation.
I couldn‘t figure out how to format the „offset“-parameter of the forward function of the deformable layer.
You‘ve mentioned in the README that currently only your core part of the code is published. Is the deformable layer currently ready-to-use?
Would be glad to hear from you!
Best regards
Moritz
The text was updated successfully, but these errors were encountered: