Thin wrapper for HPF

Thin Python wrapper for the non-parametric Hochbaum Pseudoflow (HPF) min-cut/max-flow algorithm. The original source code by Bala Chandran and Dorit S. Hochbaum is availbable here. The C++ code used in this wrapper has been refractored by Patrick M. Jensen and published here.

Installation

Install from repository (requires Cython):

git clone https://github.com/Skielex/thinhpf
cd thinhpf
pip install .

Package will be made available on PyPI later.

Tiny example

import thinhpf

hpf = thinhpf.hpf()

# Add s and t.
next_node_id = hpf.add_node(2)

def offset(n):
    return 2 + n

s = 0
t = 1

hpf.set_source(0)
hpf.set_sink(1)

# Number of nodes to add.
nodes_to_add = 2

# Add two nodes.
next_node_id = hpf.add_node(nodes_to_add)

# Add edges.
hpf.add_edge(s, offset(0), 5)  # s     --5->   n(0)
hpf.add_edge(offset(0), t, 1)  # n(0)  --1->   t
hpf.add_edge(offset(1), t, 3)  # n(1)  --3->   t
hpf.add_edge(offset(0), offset(1), 2)  # n(0)  --2->   n(1)
hpf.add_edge(offset(1), offset(0), 1)  # n(1)  --1->   n(0)
# Find maxflow/cut hpf.
hpf.mincut()
flow = hpf.compute_maxflow()

for n in range(nodes_to_add):
    segment = hpf.what_label(n)
    print('Node %d has label %d.' % (n, segment))
# Node 0 has label 0.
# Node 1 has label 1.

print('Flow: %s' % flow)
# Maximum flow: 3

Related repositories

• Hochbaum Group on GitHub
• Original source website
• slgbuilder Python package (CVPR 2020)
• shrdr Python package (ICCV 2021)
• thinqpbo Python package
• thinmaxflow Python package
• C++ implementations of max-flow/min-cut algorithms

License

The original C code by Bala Chandran and Dorit S. Hochbaum and thereby the content of hpf.h (previously pseudo.c) is published under an academic license (see LICENSE file). More information on the original website.