Skip to content

sciai-lab/pytorch-LearnedRandomWalker

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

pytorch-LearnedRandomWalker

Implementation of the LearnedRandomWalker module as described in:

Data processing:

The results reported in the paper are based on a modified version of the CREMI challenge dataset.

The following processing have been performed:

  • The raw and labels have been cropped to by 2 pixels in the x/y plane to to avoid potential misalignments during the upsampling and downsampling.
  • The slice 0 of the labels is ignored because the UNet used in the experiments uses 3 z-slices as input.
  • Some instances are connected in 3D but are not visually connected in 2D, therefore a slice-by-slice relabeling is performed.
  • Groundtruth segments smaller than 64 pixels in the x/y plane are merged with the surrounding segments using the watershed algorithm.
  • Groundtruth slices corrupted or with extreme artifacts are ignored in the testing and training. The following slices are removed from the test set: Cremi B: 44, 45, 15, 16, Cremi C: 14.
  • The first 50 valid slices from each CREMI volume (A, B, C) are used for testing. The remaining valid slices are used for training.

The final 150 test set slices (with all above mentioned modifications): https://heibox.uni-heidelberg.de/published/cvpr2019_lrw/

Additionally the repository contains seeds, learned RW segmentation, standard WS segmentation, standard RW segmentation and the CNN predictions.

Evaluation:

In the evaluation directory you can find all instruction to reproduce the results in the manuscript and the evaluation script used.

Cite:

@inproceedings{cerrone2019,
  title={End-to-end learned random walker for seeded image segmentation},
  author={Cerrone, Lorenzo and Zeilmann, Alexander and Hamprecht, Fred A},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  pages={12559--12568},
  year={2019}
}

About

Pytorch implementation of the LearnedRandomWalker module

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages