mirror of
https://github.com/wassname/scikit-image.git
synced 2026-06-28 02:01:40 +08:00
François Boulogne
64 lines
1.9 KiB
Plaintext
64 lines
1.9 KiB
Plaintext
========================
|
|
How to parallelize loops
|
|
========================
|
|
|
|
In image processing, we frequently apply the same algorithm
|
|
on a large batch of images. In this paragraph, we propose to
|
|
use `joblib <https://pythonhosted.org/joblib/>`_ to parallelize
|
|
loops. Here is an example of such repetitive tasks:
|
|
|
|
.. code-block:: python
|
|
|
|
from skimage import data, color, util
|
|
from skimage.restoration import denoise_tv_chambolle
|
|
from skimage.feature import hog
|
|
|
|
def task(image):
|
|
"""
|
|
Apply some functions and return an image.
|
|
"""
|
|
image = denoise_tv_chambolle(image[0][0], weight=0.1, multichannel=True)
|
|
fd, hog_image = hog(color.rgb2gray(image), orientations=8,
|
|
pixels_per_cell=(16, 16), cells_per_block=(1, 1),
|
|
visualise=True)
|
|
return hog_image
|
|
|
|
|
|
# Prepare images
|
|
hubble = data.hubble_deep_field()
|
|
width = 10
|
|
pics = util.view_as_windows(hubble, (width, hubble.shape[1], hubble.shape[2]), step=width)
|
|
|
|
To call the function ``task`` on each element of the list ``pics``, it is
|
|
usual to write a for loop. To measure the execution time of this loop, you can
|
|
use ipython and measure the execution time with ``%timeit``.
|
|
|
|
.. code-block:: python
|
|
|
|
def classic_loop():
|
|
for image in pics:
|
|
task(image)
|
|
|
|
|
|
%timeit classic_loop()
|
|
|
|
Another equivalent way to code this loop is to use a comprehension list which has the same efficiency.
|
|
|
|
.. code-block:: python
|
|
|
|
def comprehension_loop():
|
|
[task(image) for image in pics]
|
|
|
|
%timeit comprehension_loop()
|
|
|
|
``joblib`` is a library providing an easy way to parallelize for loops once we have a comprehension list.
|
|
The number of jobs can be specified.
|
|
|
|
.. code-block:: python
|
|
|
|
from joblib import Parallel, delayed
|
|
def joblib_loop():
|
|
Parallel(n_jobs=4)(delayed(task)(i) for i in pics)
|
|
|
|
%timeit joblib_loop()
|