Pronounced water-zed. A simple watershed and region agglomeration library for affinity graphs.

Based on the watershed implementation of Aleksandar Zlateski and Chandan Singh.

pip install waterz

Install c++ dependencies:

# linux
sudo apt install libboost-dev
# macos
brew install boost

Then

# make and activate env then:
pip install -e .

or

uv sync

import waterz
import numpy as np

# affinities is a [3,depth,height,width] numpy array of float32
affinities = ...

thresholds = [0, 100, 200]

segmentations = waterz.agglomerate(affinities, thresholds)

upgrade the version number in the pyproject.toml file, then

git tag v0.9.5
git push origin v0.9.5