A thin QPBO wrapper for Python
Project description
Thin wrapper for QPBO
Thin Python wrapper for a modified version of the quadratic pseudo-Boolean optimization (QPBO) algorithm by Vladimir Kolmogorov. The original source code by Vladimir Kolmogorov availbable at http://pub.ist.ac.at/~vnk/software.html. This wrapper uses a modified version with support for larger graphs and slightly lower memory usage. See submodule repository for more details.
QPBO vs. Maxflow
While the QPBO algorithm performs a s-t graph cut similar to Maxflow, it allows for non-submodular energy terms, which Maxflow doesn't. Amongst other things, this allows QPBO to solve optimization problems with exclusions terms, which can be very usefull. The graph constructed by the QPBO implementation is twice the size of a Maxflow graph for an equivalent problem. Thus, QPBO uses more memory and is slightly slower than Maxflow.
Installation
Install package using pip install thinqpbo
or clone this repository (including submodule). Building the package requires Cython.
Graph types
Currently, there are three different types of graphs: QPBOInt
, QPBOFloat
and QPBODouble
. The only difference is the underlying datatypes used for the edge capacities in the graph. For stability, it is recommended to use QPBOInt
for integer capacities and QPBODouble
for floating point capacities. However, in some cases, it maybe be favourable to use QPBOFloat
to reduce memory consumption.
Advanced features (QPBO-P and QPBO-I)
The QPBO implementation has a few advanced extensions known as QPBO-P and QPBO-I. Currently, not all advanced functions have been wrapped. If you need to use features of the QPBO C++ library that are not wrapped by thinqpbo
, please let me know by creating an issue on GitHub.
Tiny example
import thinqpbo as tq
# Create graph object.
graph = tq.QPBOInt()
# Number of nodes to add.
nodes_to_add = 2
# Add two nodes.
first_node_id = graph.add_node(nodes_to_add)
# Add edges.
graph.add_unary_term(0, 0, 5) # E1(0) = 5, s --5-> n(0)
graph.add_unary_term(0, 1, 0) # E0(0) = 1, n(0) --1-> t
graph.add_unary_term(1, 5, 0) # E0(1) = 5, n(1) --5-> t
graph.add_pairwise_term(0, 1, 0, 7, 0, 4) # E01(0,1) = 7, n(0) --7-> n(1)
# E11(0,1) = 4, Not possible with standard Maxflow
# Find maxflow/cut graph.
graph.solve()
graph.compute_weak_persistencies()
twice_energy = graph.compute_twice_energy()
for n in range(nodes_to_add):
segment = graph.get_label(n)
print('Node %d has label %d.' % (n, segment))
# Node 0 has label 0.
# Node 1 has label 0.
print('Twice energy/flow: %s' % twice_energy)
# Twice energy/flow: 12
License
As the QPBO implementation is distributed under the GPLv3 license, so is this package.
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distributions
File details
Details for the file thinqpbo-0.1.4.tar.gz
.
File metadata
- Download URL: thinqpbo-0.1.4.tar.gz
- Upload date:
- Size: 112.9 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.2.0 pkginfo/1.6.1 requests/2.24.0 setuptools/49.2.1 requests-toolbelt/0.9.1 tqdm/4.51.0 CPython/3.9.0
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | bc0475bb93fa37e0af64efbd3a8c487941b5aa671e668028d144e28453887846 |
|
MD5 | fdbdbff71077a8e8ceacfc8c2911c97e |
|
BLAKE2b-256 | e982f80001cb54d3d6d7203943565a766c2e85b67b75084c458d8568b555c92d |
File details
Details for the file thinqpbo-0.1.4-cp39-cp39-win_amd64.whl
.
File metadata
- Download URL: thinqpbo-0.1.4-cp39-cp39-win_amd64.whl
- Upload date:
- Size: 75.4 kB
- Tags: CPython 3.9, Windows x86-64
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.4.2 importlib_metadata/4.8.1 pkginfo/1.7.1 requests/2.26.0 requests-toolbelt/0.9.1 tqdm/4.62.3 CPython/3.10.0
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | a41b30458f98e84c7c451a3564abddf6bed655f3bb0751ace89f56bd7996f664 |
|
MD5 | 8cb9fa61e786988b396fc330105163ff |
|
BLAKE2b-256 | c4870c28ace1026fd099b6267299ab5c940f18be30763d52cdc03fbafc37101d |
File details
Details for the file thinqpbo-0.1.4-cp39-cp39-macosx_10_14_x86_64.whl
.
File metadata
- Download URL: thinqpbo-0.1.4-cp39-cp39-macosx_10_14_x86_64.whl
- Upload date:
- Size: 100.1 kB
- Tags: CPython 3.9, macOS 10.14+ x86-64
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.4.2 importlib_metadata/4.8.1 pkginfo/1.7.1 requests/2.26.0 requests-toolbelt/0.9.1 tqdm/4.62.3 CPython/3.10.0
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 85590bbdd4dfbe5d09008278a0aa9423639576c81e84fb074b50e6eff6bbcb8f |
|
MD5 | ad36113702d1145310bb8dad1639f250 |
|
BLAKE2b-256 | c2931b456df8218e9e014e321927d38b1f43fb5d594937ff219e7a03443c8e8e |
File details
Details for the file thinqpbo-0.1.4-cp38-cp38-win_amd64.whl
.
File metadata
- Download URL: thinqpbo-0.1.4-cp38-cp38-win_amd64.whl
- Upload date:
- Size: 75.4 kB
- Tags: CPython 3.8, Windows x86-64
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.4.2 importlib_metadata/4.8.1 pkginfo/1.7.1 requests/2.26.0 requests-toolbelt/0.9.1 tqdm/4.62.3 CPython/3.10.0
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 16c28700cfbed7211555886e8fac4a3b56774361312fcb4b7df4898d023ad0f0 |
|
MD5 | bff4d63e80f9f424b0b016b2e194ac10 |
|
BLAKE2b-256 | 0ee13816237ed1a7468b8f1204e11c5eff9b12de8146c1a053c028fccd1bd76c |
File details
Details for the file thinqpbo-0.1.4-cp38-cp38-macosx_10_14_x86_64.whl
.
File metadata
- Download URL: thinqpbo-0.1.4-cp38-cp38-macosx_10_14_x86_64.whl
- Upload date:
- Size: 99.9 kB
- Tags: CPython 3.8, macOS 10.14+ x86-64
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.4.2 importlib_metadata/4.8.1 pkginfo/1.7.1 requests/2.26.0 requests-toolbelt/0.9.1 tqdm/4.62.3 CPython/3.10.0
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 19ec9e7e1378037445846c7c97de92c319b691ae2695fda45f821d7ed97016cf |
|
MD5 | 5ee217e51530e1b290226e272b4bb351 |
|
BLAKE2b-256 | 15f4c851497da906f53d85c10e9be192fecb2001896cd7d1ea937df46d0c3dad |
File details
Details for the file thinqpbo-0.1.4-cp37-cp37m-win_amd64.whl
.
File metadata
- Download URL: thinqpbo-0.1.4-cp37-cp37m-win_amd64.whl
- Upload date:
- Size: 74.9 kB
- Tags: CPython 3.7m, Windows x86-64
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.2.0 pkginfo/1.6.1 requests/2.24.0 setuptools/49.2.1 requests-toolbelt/0.9.1 tqdm/4.51.0 CPython/3.9.0
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 8e7a9b217022e934a26260acd85a23e351911fbe2bfd823b78e030dcb5776fe3 |
|
MD5 | d3033288d1deba757f469641cdd22f67 |
|
BLAKE2b-256 | ab99965714525cbbfb1b64fe327fb30d11557220cfc58170315770d8693c6bfb |
File details
Details for the file thinqpbo-0.1.4-cp37-cp37m-macosx_10_14_x86_64.whl
.
File metadata
- Download URL: thinqpbo-0.1.4-cp37-cp37m-macosx_10_14_x86_64.whl
- Upload date:
- Size: 102.7 kB
- Tags: CPython 3.7m, macOS 10.14+ x86-64
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.2.0 pkginfo/1.6.1 requests/2.24.0 setuptools/49.2.1 requests-toolbelt/0.9.1 tqdm/4.51.0 CPython/3.9.0
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | c415922b4c3eddd6ce2422cf8538057428dc0cf0fc3e1d20fa201e8a4078f84e |
|
MD5 | 8c334e37a34fba328ef96c159f830862 |
|
BLAKE2b-256 | a0a0b63183c603c7ded17efc345dd8fad6aa3dc23b89b6cd09ddd481707e65d3 |
File details
Details for the file thinqpbo-0.1.4-cp36-cp36m-win_amd64.whl
.
File metadata
- Download URL: thinqpbo-0.1.4-cp36-cp36m-win_amd64.whl
- Upload date:
- Size: 75.0 kB
- Tags: CPython 3.6m, Windows x86-64
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.2.0 pkginfo/1.6.1 requests/2.24.0 setuptools/49.2.1 requests-toolbelt/0.9.1 tqdm/4.51.0 CPython/3.9.0
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 700c82f78b5a1b88fec2e226e1c19529ef95224fec060b86a2af5a38d72c9142 |
|
MD5 | 5b6da3e38eb436cfcdcd3b746534091e |
|
BLAKE2b-256 | 0eff88430eab91ef38e20934f38fb21126698bf34b7718f1b639eca77253ffc8 |
File details
Details for the file thinqpbo-0.1.4-cp36-cp36m-macosx_10_14_x86_64.whl
.
File metadata
- Download URL: thinqpbo-0.1.4-cp36-cp36m-macosx_10_14_x86_64.whl
- Upload date:
- Size: 102.8 kB
- Tags: CPython 3.6m, macOS 10.14+ x86-64
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.2.0 pkginfo/1.6.1 requests/2.24.0 setuptools/49.2.1 requests-toolbelt/0.9.1 tqdm/4.51.0 CPython/3.9.0
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 64f679fa873fa05bb9e9f81bc15376be8baedf49df32de7846afb6cf11753c8f |
|
MD5 | 6ef0c82831c73af5d089b22e9bde701a |
|
BLAKE2b-256 | 07750d181bcefe03b70fea3cd42d0247c4032bfc2f3f636535ed86dfdc5090a4 |
File details
Details for the file thinqpbo-0.1.4-cp27-cp27m-macosx_10_14_x86_64.whl
.
File metadata
- Download URL: thinqpbo-0.1.4-cp27-cp27m-macosx_10_14_x86_64.whl
- Upload date:
- Size: 99.1 kB
- Tags: CPython 2.7m, macOS 10.14+ x86-64
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.2.0 pkginfo/1.6.1 requests/2.24.0 setuptools/49.2.1 requests-toolbelt/0.9.1 tqdm/4.51.0 CPython/3.9.0
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 41f3ba1cc4b27a95fdd8a60f56ad874f112d4e124c44e82114b0b34dd95c43fd |
|
MD5 | 4654685bee74f225e43a364f211bf4dc |
|
BLAKE2b-256 | b475c5d18931f6633f3b633912764e4ccabbbc3a90ed491afb00011c689f87bb |