Pronounced water-zed. A simple watershed and region agglomeration library for affinity graphs.
Incorporate functions from Mala_v2.zip from http://cremi.org for better segmentation result
conda create -n zw python==3.8 numpy cython
pip install --editable .
import waterz
import numpy as np
# affinities is a [3,depth,height,width] numpy array of float32
affinities = ...
# evaluation: vi/rand
seg_gt = None
aff_thresholds = [0.005, 0.995]
seg_thresholds = [0.1, 0.3, 0.6]
seg = waterz.waterz(aff, seg_thresholds, merge_function='aff50_his256',
aff_threshold=aff_thresholds, gt=seg_gt)