Mailman 3 December 2012 - scikit-image

import speedup
by Steven Silvester Dec. 14, 2012

Dec. 14, 2012

Hey all, I noticed that loading skimage seemed to take a bit longer than I'd like, so I profiled it. On my machine, it takes 1.6 seconds to import skimage, 0.4 of which was from numpy, and 1.2 of which was from loading nose in skimage/__init__.py. This simple patch brings my import time down to ~0.4 seconds, since it only imports nose when the test function is called. diff --git a/skimage/__init__.py b/skimage/__init__.py index 1ac121b..f3e5e88 100644 --- a/skimage/__init__.py +++ b/skimage/__init__.py @@ -84,8 +84,9 @@ def _setup_test(verbose=False): return f -test = _setup_test() -test_verbose = _setup_test(verbose=True) +test = lambda : _setup_test()() +test_verbose = lambda : _setup_test(verbose=True)() +test_verbose.__doc__ = test.__doc__ = 'Invoke the skimage test suite.' def get_log(name=None):

3 2

Re: Stable vs Dev version, document image enhancement
by Stéfan van der Walt Dec. 14, 2012

Dec. 14, 2012

On Thu, Dec 13, 2012 at 8:54 PM, Tony Yu <tsyu80(a)gmail.com> wrote: > It might be a good idea for us to point to the stable version instead of the > development version by default. The development gallery is useful as an advertisement, but we could maybe make the link to the other versions clearer? Stéfan

1 0

Re: Stable vs Dev version, document image enhancement
by Josh Warner Dec. 13, 2012

Dec. 13, 2012

Hi Parand, Welcome! Most of the gallery examples work in the current release (0.7.2 I believe), but the rank filter is in a PR that was merged recently, and is presently only in the dev version. You may want the functionality in the rank filter for OCR work, where fast local operations like the otsu threshold could be very beneficial. I'll let other people answer the questions about specific references as I'm on the road, but I think most of what you need to do is available in scikit-image. Josh Warner On Wednesday, December 12, 2012 7:00:24 PM UTC-6, Parand Darugar wrote: > > Hello, > > I'm new to scikit-image, having mostly used PIL and imagemagick to date. > Looks very promising, thanks for putting it together. > > A few newbie questions: > > - It seems the examples in the gallery are based on the dev version as > opposed to the stable version? Therefore I assume it's recommended to use > the dev version instead of the "pip install" version? I just tried the > latest from git and am getting some errors (see below) - which version > should I be using? > > - My initial goal is image/document enhancement and cleanup in order to > improve OCR. The images are primarily captured via cell phone cameras and > scanners, so I'm looking to normalize and enhance the color, remove noise, > and straighten them. Any pointers to code, papers, or algorithms that may > be of interest would be appreciated. > > -- Error message -- > >>> import skimage.filter.rank as rank > Traceback (most recent call last): > File "<stdin>", line 1, in <module> > File "skimage/filter/__init__.py", line 2, in <module> > from .ctmf import median_filter > File "skimage/filter/ctmf.py", line 15, in <module> > from . import _ctmf > ImportError: cannot import name _ctmf > > Much thanks, > > Parand >

2 1

canvastools subpackage for scikit-image viewer
by Tony Yu Dec. 13, 2012

Dec. 13, 2012

About a month-or-so ago, I promised I would start work on selection tool for the image viewer. I finally got around to it. I ended up doing a pretty significant overhaul, but I think it was worth it. See PR # 390 <https://github.com/scikit-image/scikit-image/pull/390> I'll copy my PR message below, just for completeness: Major overhaul of tools for interacting with the scikit-image viewer. * Add skimage.viewer.canvastools subpackage with line and rectangle selection tools * Refactor line profile plugin to use line tool * Fix viewer canvas * Fix blitting behavior * New plugins (ColorHistogram, Measure, Crop) To get a feel for the changes, see new examples in `viewer_examples/plugins` located in the root directory of package. Cheers, -Tony

1 0

Re: Hello, everyone
by caohan Dec. 7, 2012

Dec. 7, 2012

On 12/07/2012 05:06 AM, StÃ¯Â¿Â½fan van der Walt wrote: > Hi there, > > On Wed, Dec 5, 2012 at 5:58 PM, Han Cao <simonda72819(a)gmail.com> wrote: >> I am new guy to this group. I am a student who want to get some development >> experiences to image processing products. So I choose scikit-image. I want >> to contribute to this product. >> I hope to know is there any products using this library now? Thanks a lot if >> anyone answer me? > Welcome, I hope you find the project useful! > > Are you asking about commercial products specifically, or research > projects in general? Scikit-image has found all sorts of interesting > applications, ranging from astronomy to biology to computer vision. I > think it would be a good idea to gather more such "user stories"! > > StÃ¯Â¿Â½fan > Thank you for replying me, I am asking about commercial products, such as Matlab? Hank Cao

1 0

Re: Hello, everyone
by Stéfan van der Walt Dec. 6, 2012

Dec. 6, 2012

Hi there, On Wed, Dec 5, 2012 at 5:58 PM, Han Cao <simonda72819(a)gmail.com> wrote: > I am new guy to this group. I am a student who want to get some development > experiences to image processing products. So I choose scikit-image. I want > to contribute to this product. > I hope to know is there any products using this library now? Thanks a lot if > anyone answer me? Welcome, I hope you find the project useful! Are you asking about commercial products specifically, or research projects in general? Scikit-image has found all sorts of interesting applications, ranging from astronomy to biology to computer vision. I think it would be a good idea to gather more such "user stories"! Stéfan

1 0