The segraph library provides modules for creating graphs from SLIC segments. This can be used with PyStruct library for image segmentation using CRF.
Link to GitHub RepoThese instructions will get you a copy of the project up and running on your local machine for development and testing purposes. See examples to see a quick example.
You will need to have NumPY installed on your system. The segraph library was built with Python 2.7, keeping in mind the stability with Python 3+, but it is not guranteed.
pip install numpy
There are multiple ways to install segraph on your system:
segraph is now available at https://pypi.python.org/pypi/segraph/0.5
1. Download the tar/zip from https://pypi.python.org/pypi/segraph/0.5
2. Move the package to your desired location / python version, and unzip the archive.
Optionally, if you have a linux-based machine (Ubuntu/OSX):
tar xvzf segraph-0.x.tar.gz -C /path/to/desireddirectory
3. Migrate to the segraph folder, and run
python setup.py install
pip install segraph
To upgrade,
pip install --upgrade segraph
segraph can be very helpful for creating graphs from SLIC segmented images (superpixels). Here is an example usage:
from skimage.segmentation import slic
from skimage.util import img_as_float
from skimage import io as skimageIO
from segraph import create_graph
import numpy as np
image = img_as_float(skimageIO.imread("segraph/data/flowers.png"))
segments = slic(image, n_segments=500, sigma=1.0)
# Create graph of superpixels
vertices, edges = create_graph(segments)
# Compute centers:
gridx, gridy = np.mgrid[:segments.shape[0], :segments.shape[1]]
centers = dict()
for v in vertices:
centers[v] = [gridy[segments == v].mean(), gridx[segments == v].mean()]
segraph can be used with PyStruct library for image segmentation using CRF.
You are welcome to send a pull-request.
- Abhinandan Dubey - @alivcor - Human Interaction Lab, Stony Brook University
This project is licensed under the GNU General Public License v3 - see the LICENSE.md file for details
