ryany1994/mutex-watershed

The mutex watershed for image segmentation.

Mutex Watershed

The Mutex Watershed algorithm for efficient segmentation without seeds. For the corresponding publication see: http://openaccess.thecvf.com/content_ECCV_2018/html/Steffen_Wolf_The_Mutex_Watershed_ECCV_2018_paper.html

Installation

On Unix (Linux, OS X)

• create conda env with xtensor:
• conda create -n mws -c conda-forge xtensor-python
• activate the env:
• source activate mws
• clone this repository, enter it and install:
• python setup.py install

ISBI Experiments

To reproduce the ISBI experiments, go to the experiments/isbi folder and run the isbi_experiments script: python isbi_experiments.py /path/to/raw.h5 /path/to/affinities.h5 /path/to/res_folder --algorithms mws

You will need vigra as additional dependency: conda install -c conda-forge vigra

You can also reproduce the baseline results by specifying further algorithms. Note that most of these will need the https://github.com/constantinpape/cremi_tools repository and further dependencies specified there.

You can obtain the data from: https://hcicloud.iwr.uni-heidelberg.de/index.php/s/6LuE7nxBN3EFRtL

Training affinity networks

To train an affinity network for isbi, you can use the scripts in experiments/training. You will also need inferno and neurofire