waterz

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

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

Installation

python setup.py install

Requires numpy and cython.

Usage

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)