Mailman 3 June 2018 - scikit-image

Re: [scikit-image] rescale with nearest neighbour gives unpredictable results
by Juan Nunez-Iglesias 12 Jun '18

12 Jun '18

Hi Jon, Sorry for the long delay in responding! It has been a very busy 6 weeks for the skimage dev, and this looks to continue. Anyway, my first impression is that your error is very similar to this issue: https://github.com/scikit-image/scikit-image/issues/2629 for which we have a proposed fix here: https://github.com/scikit-image/scikit-image/pull/2648 As you can see from that discussion, it's unclear whether we want to merge this. From my perspective, it falls more on the side of "it's unavoidable". As Greg Lee commented, his hacky solution doesn't work in all cases, and to me it seems that it sets us up for other errors later. Let us know what you think. The above is just my opinion and skimage is community-driven. Regarding your specific problem, though, I recommend you look at `downscale_local_mean`, which i believe is the more correct way to downscale your data. Juan. On Wed, Apr 25, 2018, at 7:36 PM, Jon Morris wrote: > Hello all, > > Just had a query about transform.rescale (and by extension, warp), > which is giving inconsistent results between different platforms. I > wondered if this was a bug or was unavoidable.> > Version info: > numpy==1.13.0 > scikit-image==0.13.1 > scipy==0.19.1 > > Platforms: > Scientific linux 7.4 and Windows 10 > > So, I have a raster I'm rescaling from 5m resolution to 10m: > In [35]: import numpy as np > In [36]: input_array=np.array([[ 30.7213501 , 30.73872986, > 30.77840255, 30.79360602],> ...: [ 30.40055123, 30.39305344, 30.40674613, > 30.39675925],> ...: [ 30.70790547, 30.67351216, 30.7357262 , > 30.70840534],> ...: [ 30.3960635 , 30.33706349, 30.38129123, > 30.34338268]])> > In [37]: from skimage.transform import rescale > In [38]: > rescale(input_array,0.5,mode='symmetric',preserve_range=True,order=0)> Out[38]: > array([[ 30.39305344, 30.77840255], > [ 30.33706349, 30.70840534]]) > > When I try the same thing on a different linux server (same distro and > version), the output is different.> Out[6]: > array([[ 30.39305344, 30.40674613], > [ 30.33706349, 30.70840534]]) > > Over on windows, I get a different answer again. > Out[125]: > array([[ 30.39305344, 30.79360602], > [ 30.33706349, 30.34338268]]) > > It looks like, when downsampling with nearest neighbour interpolation, > for each group of four pixels, one is chosen for the output raster, > but which one is chosen varies on different platforms. I stepped > through the code and the difference happens in the _warp_fast function > in _warps_cy, so I couldn't tell exactly why each value was chosen.> > Is this expected behaviour, or should the same value be chosen on each > platform when downsampling? It's messing with our unit tests, as we > need a different expected value for each platform we run the tests on.> > Thanks, > > Jon > > > > *Jon Morris* > Software Developer > > *Skype[1]* > > *T* +44 (0) 1756 799919 > www.jbarisk.com > Visit our website Follow us on Twitter > _________________________________________________ > scikit-image mailing list > scikit-image(a)python.org > https://mail.python.org/mailman/listinfo/scikit-image Links: 1. sip:jon.morris@jbarisk.com

1 0

Re: [scikit-image] 回复：Re: imagepy new version
by David Sarma 07 Jun '18

07 Jun '18

Is there any thought to porting the scikit-image project to other language communities? I'm thinking in particular of Javascript, wrt the recent developments surrounding Tensorflow.js bringing Numpy-like structures (with GPU acceleration), in a form that's maybe a little more stable than the typical Javascript library. There's a breakdown of some of the Javascript equivalents of Python libraries and features here: https://beta.observablehq.com/@tmcw/javascript-replacements-for-python-data… . While Javascript as of a few years ago I found to be a somewhat frustrating experience (though its infrastructure very useful for prototyping), it appears to have matured recently (possibly as a result of Web Assembly). In relation to image processing/scientific computing in particular, I noticed for example OpenCV.js <https://docs.opencv.org/3.4.1/df/d0a/tutorial_js_intro.html> and VTK.js <https://kitware.github.io/vtk-js/>. Making use of some of these could potentially help address some of the cross-platform compatibility issues. On Tue, Jun 5, 2018 at 7:48 PM, Juan Nunez-Iglesias <jni(a)fastmail.com> wrote: > On Wed, Jun 6, 2018, at 7:08 AM, Stefan van der Walt wrote: > > A different consideration is whether there is risk of legal trouble: > > that risk is probably quite low, with Pyside2 / Qt at least. > > I think this is true. If Qt intended to sue open source BSD projects using > Qt, Matplotlib would not exist. Nor Spyder. (There's probably other > examples.) > > For the scikit-image project, the other considerations are important. > Scikit-image is intended to sit near the base of a project, imported as a > library and mixed with other code to create a new product. For a GUI, in my > opinion, the GUI itself is the end product, and I'm not too concerned about > it "technically" being GPL. > > Juan. > _______________________________________________ > scikit-image mailing list > scikit-image(a)python.org > https://mail.python.org/mailman/listinfo/scikit-image > -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Phone 917 375 8730 Email david(a)davidsarma.org Web http://www.davidsarma.org

1 0

2018 John Hunter Excellence in Plotting Contest
by Nelle Varoquaux 06 Jun '18

06 Jun '18

Hello everyone, Sorry about the cross-posting. There's a couple more days to submit to the John Hunter Excellence in Plotting Competition! If you have any scientific plot worth sharing, submit an entry before June 8th. For more information, see below. Thanks, Nelle In memory of John Hunter, we are pleased to be reviving the SciPy John Hunter Excellence in Plotting Competition for 2018. This open competition aims to highlight the importance of data visualization to scientific progress and showcase the capabilities of open source software. Participants are invited to submit scientific plots to be judged by a panel. The winning entries will be announced and displayed at the conference. John Hunter’s family and NumFocus are graciously sponsoring cash prizes for the winners in the following amounts: - 1st prize: $1000 - 2nd prize: $750 - 3rd prize: $500 - Entries must be submitted by June, 8th to the form at https://goo.gl/forms/7q86zgu5OYUOjODH3 <https://goo.gl/forms/7q86zgu5OYUOjODH3>. - Winners will be announced at Scipy 2018 in Austin, TX. - Participants do not need to attend the Scipy conference. - Entries may take the definition of “visualization” rather broadly. Entries may be, for example, a traditional printed plot, an interactive visualization for the web, or an animation. - Source code for the plot must be provided, in the form of Python code and/or a Jupyter notebook, along with a rendering of the plot in a widely used format. This may be, for example, PDF for print, standalone HTML and Javascript for an interactive plot, or MPEG-4 for a video. If the original data can not be shared for reasons of size or licensing, "fake" data may be substituted, along with an image of the plot using real data. - Each entry must include a 300-500 word abstract describing the plot and its importance for a general scientific audience. - Entries will be judged on their clarity, innovation and aesthetics, but most importantly for their effectiveness in communicating a real-world problem. Entrants are encouraged to submit plots that were used during the course of research or work, rather than merely being hypothetical. - SciPy reserves the right to display any and all entries, whether prize-winning or not, at the conference, use in any materials or on its website, with attribution to the original author(s). SciPy John Hunter Excellence in Plotting Competition Co-Chairs Thomas Caswell Michael Droettboom Nelle Varoquaux

1 0

Re: [scikit-image] 回复：Re: imagepy new version
by Stefan van der Walt 05 Jun '18

05 Jun '18

On Tue, 05 Jun 2018 11:32:19 +1000, Juan Nunez-Iglesias wrote: > My understanding is that certain PyQt wrappers actually get around the > GPL requirement, allowing you to link to it from BSD-licensed code. > > (Which is how things like Matplotlib and Spyder are BSD/MIT.) > But, it's true: none of the Python GUI frameworks are perfect. This issue has been discussed on the list before, and I am still unconvinced that you can legally "get around" the GPL requirements. It's a murky issue, complicated also by Qt's dual licensing model. My understanding is that you may distribute your code under whichever license you choose, but that the binary product that results from importing e.g. matplotlib + qt together is governed by the GPL. The "workaround" is thus to postpone the instant at which linking occurs (given that "import" is Python's way of linking). What are the implications of doing so? Could a lawayer argue that, in spirit, the GPL is being violated? A different consideration is whether there is risk of legal trouble: that risk is probably quite low, with Pyside2 / Qt at least. Yet another issue is our responsibility to our users. With patents, e.g., our decision has been not to include any encumbered algorithms, because there aren't very many good ways of protecting our users against them (you copy/paste some skimage code from StackOverflow, run it, and all of a sudden you've applied a patented algorithm on your data—yikes!). It is not clear how we should warn our users in this case, or what exactly that warning should say. Best regards, Stéfan

2 1

Re: [scikit-image] 回复：Re: imagepy new version
by Juan Nunez-Iglesias 05 Jun '18

05 Jun '18

Thanks Mark, those links are very useful! Pretty excited about the new PySide2 (or Qt for Python as the case may be), as it seems that Python may finally be considered a first-class citizen by the Qt devs. On Tue, Jun 5, 2018, at 2:22 PM, Mark Harfouche wrote: > I have to agree with Juan. > > I've been trying to keep up with Jupyter Notebooks and they just don't > seem there yet.> I brought this up in an earlier post, but visbrain seems to be a great > proof of concept as to what is possible with Qt and Python> http://visbrain.org/ > > Also see PySide2 coming soon: > http://blog.qt.io/blog/2018/04/13/qt-for-python-is-coming-to-a-computer-nea…> LGPL gives you the power of the library with minimal requirements. > Maybe LGPL is enough for you?

1 0

Re: [scikit-image] 回复：Re: imagepy new version
by Mark Harfouche 05 Jun '18

05 Jun '18

I have to agree with Juan. I've been trying to keep up with Jupyter Notebooks and they just don't seem there yet. I brought this up in an earlier post, but visbrain seems to be a great proof of concept as to what is possible with Qt and Python http://visbrain.org/ Also see PySide2 coming soon: http://blog.qt.io/blog/2018/04/13/qt-for-python-is-coming-to-a-computer-nea… LGPL gives you the power of the library with minimal requirements. Maybe LGPL is enough for you? On Mon, Jun 4, 2018 at 6:32 PM Juan Nunez-Iglesias <jni(a)fastmail.com> wrote: > On Tue, Jun 5, 2018, at 2:21 AM, imagepy(a)sina.com wrote: > > Sorry，ImagePy is a ui framework based on wxpython, which can not install > with pip on Linux. You need download the whl acrodding to your Linux > system. > > > Ah, that's a shame. I'm on 18.04, so I don't even know if the 16.04 wheel > would work. I suppose not. > > wx's official said, wx is a native ui framework, so cannot fit different > Linux and gtk version. I had pull a issue to solve it here, and someone had > started it. > > > Yikes. > > About the future, I think web is the best direction. I choose wx, not qt, > because qt has a gpl license, and both qt and wx is very heavy. and I am > not good at js. But now the jupyterlab is just ok, I will start this work > sometime when the wx version is stable. > > > I actually disagree with this. I was having a discussion with a friend > about how old computers (1980s) actually felt more responsive than current > ones, because the GUIs were not so wasteful of processing cycles. I think > the trend to web UIs is worsening that situation. Web apps are way less > responsive than well-implemented native apps. (And they are a leaky > abstraction.) > > My understanding is that certain PyQt wrappers actually get around the GPL > requirement, allowing you to link to it from BSD-licensed code. (Which is > how things like Matplotlib and Spyder are BSD/MIT.) > > But, it's true: none of the Python GUI frameworks are perfect. > > Juan. > _______________________________________________ > scikit-image mailing list > scikit-image(a)python.org > https://mail.python.org/mailman/listinfo/scikit-image >

1 0

回复：Re: imagepy new version
by imagepy＠sina.com 05 Jun '18

05 Jun '18

Sorry，ImagePy is a ui framework based on wxpython, which can not install with pip on Linux. You need download the whl acrodding to your Linux system. wx's official said, wx is a native ui framework, so cannot fit different Linux and gtk version. I had pull a issue to solve it here, and someone had started it. About the future, I think web is the best direction. I choose wx, not qt, because qt has a gpl license, and both qt and wx is very heavy. and I am not good at js. But now the jupyterlab is just ok, I will start this work sometime when the wx version is stable. BestYXDagon ----- 原始邮件 ----- 发件人："Juan Nunez-Iglesias" <jni(a)fastmail.com> 收件人：scikit-image(a)python.org 主题：Re: [scikit-image] imagepy new version 日期：2018年06月01日 13点11分 Hi Yan, Thanks for writing! And congrats on the paper! Unfortunately, my ImagePy install is failing. I was super-happy to discover that conda install conda-forge::imagepy works, but then: $ python -m imagepy Traceback (most recent call last): File "/home/jni/miniconda3/envs/imagepy/lib/python3.6/runpy.py", line 183, in _run_module_as_main mod_name, mod_spec, code = _get_module_details(mod_name, _Error) File "/home/jni/miniconda3/envs/imagepy/lib/python3.6/runpy.py", line 142, in _get_module_details return _get_module_details(pkg_main_name, error) File "/home/jni/miniconda3/envs/imagepy/lib/python3.6/runpy.py", line 109, in _get_module_details __import__(pkg_name) File "/home/jni/miniconda3/envs/imagepy/lib/python3.6/site-packages/imagepy/__init__.py", line 3, in <module> import sys, os, wx File "/home/jni/miniconda3/envs/imagepy/lib/python3.6/site-packages/wx/__init__.py", line 17, in <module> from wx.core import * File "/home/jni/miniconda3/envs/imagepy/lib/python3.6/site-packages/wx/core.py", line 12, in <module> from ._core import * ImportError: libpng12.so.0: cannot open shared object file: No such file or directory Any hints? I tried adding the right `lib` path to LD_LIBRARY_PATH but it's still failing. I appear to have libpng 1.6 rather than 1.2... Juan. On Fri, Jun 1, 2018, at 2:27 PM, imagepy(a)sina.com wrote: Thanks for Releasing 0.14, And I want to make a introduction for ImagePy. It is a plugin framework for image process, just like imagej. I had introduced befor, and now it becomes stronger! can wraper not only numpy, but also pandas. Now it is like imagej + spss. And I had published a paper in Bioinformatics here. A snapshot attached. Welcom to have a look! Best YXDragon _______________________________________________ scikit-image mailing list scikit-image(a)python.org https://mail.python.org/mailman/listinfo/scikit-image Email had 1 attachment: imagepy.png 517k (image/png) _______________________________________________ scikit-image mailing list scikit-image(a)python.org https://mail.python.org/mailman/listinfo/scikit-image

2 1

imagepy new version
by imagepy＠sina.com 01 Jun '18

01 Jun '18

Thanks for Releasing 0.14, And I want to make a introduction for ImagePy.It is a plugin framework for image process, just like imagej. I had introduced befor, and now it becomes stronger!can wraper not only numpy, but also pandas. Now it is like imagej + spss. And I had published a paper in Bioinformatics here.A snapshot attached. Welcom to have a look! BestYXDragon

2 1