Skip to content

A baseline 3D U-Net implementation that follows the architecture of Cicek et al. as close as possible, implemented in TF2/keras

License

Notifications You must be signed in to change notification settings

wangfaofao/3DU-Net

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

3D U-Net

This repository tries to follow the architecure of Özgün Çiçek et al. 3D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotation as closely as possible. As seen in the below figure, the number of parameters in this implementation is very similar to those reported in the paper but not exactly the same. I think this difference may be due to the counting of the batch normalisation parameters.

Comparison of parameters

There has been no attempt to optimise the performance of this architecture. However, a concise, modular, class-based implementation of the 3D U-Net is provided and as such it is probably more suited as a learning/teaching resource.

About

A baseline 3D U-Net implementation that follows the architecture of Cicek et al. as close as possible, implemented in TF2/keras

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%