Mailman 3 March 2012 - scikit-image

Re: RFC: Multiple backends + speedy looping
by Stéfan van der Walt 15 Mar '12

15 Mar '12

Hi Tony On Wed, Mar 14, 2012 at 5:28 PM, Tony Yu <tsyu80(a)gmail.com> wrote: > From the looks of the implementation, it seems none of the current modules > would need to be changed unless a backend was added. And even then, it looks > simple to modify an existing function to add a backend. Does this sum it up: > > * Add `@add_backends` to function with new (optional) backend > * register backend function using `backend.register` in the `__init__` > file for the function > * Subclass tests from `BackendTester` (an additional decorator for test > functions might be good, if possible) That's right--the design is aimed to be minimally invasive. > Stefan: Do you have an idea of what needs to be done to get this merged? > After a quick look: 1) It needs to be moved from the scikits.image namespace > to skimage, 2) there are questionable changes to "filter/edges.py" and > "filter/tests/test_edges.py" in the PR, 3) implementation should be moved > from "backend/__init__.py" to something like "backend/core.py" (or I guess > we've been using the convention "backend/_backend.py"). > Also, I'm not crazy about dumping backend functions into a "backend" > subdirectory for each module. Maybe they should go in a backend-specific > subdirectory. So, for example, "filter/backend/edges_opencl.py" would move > to "filter/opencl/edges.py". You've identified the three core issues. (1) is easy to do. (2) -- I believe we should rework or remove those examples. (3) I share your sentiment--skimage.filter.opencl seems like a better place. Thanks for the valuable comments! The more I think about this, the more I think that this can be a great way of experimentally incorporating new and maturing technologies, such as numba. Even if we can just start by having the fastest convolution of all the available packages--without giving up the advantages of high-level description--I'd be very happy. Stéfan

1 0

Re: RFC: Multiple backends + speedy looping
by Tony Yu 14 Mar '12

14 Mar '12

2012/3/1 Stéfan van der Walt <stefan(a)sun.ac.za> > Hi Nadav > > On Wed, Feb 29, 2012 at 9:36 PM, Nadav Horesh <nadavh.horesh(a)gmail.com> > wrote: > > I am working on a simple frame work as a replacement for > > ndimage.genereic_filter and numeric's neighbourhood iterator (I hope it > > would be faster and easier to use). > > The goal of this PR is to support any backend (including, possibly, > the one you are writing). I'd like to provide multiple > implementations of some commonly used functions (convolution, e.g.), > so that we can have a performance based series of fallbacks, such as > GPU -> Cython -> Python. That way, algorithms can be run on any > system, but will execute most rapidly on those supporting, say, > OpenCL. Another advantage is that if, in the future, we write an even > faster implementation, the benefit is immediately available to the > user, without rewriting any code. > > Stéfan > >From the looks of the implementation, it seems none of the current modules would need to be changed unless a backend was added. And even then, it looks simple to modify an existing function to add a backend. Does this sum it up: * Add `@add_backends` to function with new (optional) backend * register backend function using `backend.register` in the `__init__` file for the function * Subclass tests from `BackendTester` (an additional decorator for test functions might be good, if possible) The rest of the changes would happen in separate files for each function implementation. If this is an accurate description, then I think the backend system would be a good addition and wouldn't really add any annoying boilerplate code. Stefan: Do you have an idea of what needs to be done to get this merged? After a quick look: 1) It needs to be moved from the scikits.image namespace to skimage, 2) there are questionable changes to "filter/edges.py" and "filter/tests/test_edges.py" in the PR, 3) implementation should be moved from "backend/__init__.py" to something like "backend/core.py" (or I guess we've been using the convention "backend/_backend.py"). Also, I'm not crazy about dumping backend functions into a "backend" subdirectory for each module. Maybe they should go in a backend-specific subdirectory. So, for example, "filter/backend/edges_opencl.py" would move to "filter/opencl/edges.py". -Tony

1 0

Numba: fast Python loops via LLVM
by Stéfan van der Walt 12 Mar '12

12 Mar '12

Hi all, Travis announced his LLVM-based compiler for Python loops at PyCon2012: https://github.com/ContinuumIO/numba I think this is an interesting option for skimage. If we integrate the multiple-backend system, we can very easily slot into this mechanism. Stéfan

1 0

Re: Imshow and Linescan - first version
by Guillaume Gay 10 Mar '12

10 Mar '12

Le 26/02/2012 21:15, Stéfan van der Walt a écrit : > On Sat, Feb 25, 2012 at 11:10 AM, Tony Yu<tsyu80(a)gmail.com> wrote: >> Plugin system >> ------ >> It might be nice to have some sort of plugin system for the image viewer. >> That way, tools like this could be implemented a bit more easily and, also, >> easily added to the viewer or ignored by the user. > At the moment, we basically have a double-viewer implementation: > simply vs fancy. So we separate out > > imshow(x) and imshow(x, fancy=True) I thought one solution was to sub-class FigureCanvas, with extra buttons, but from what I understand, it is not that easy to do this at the platform independant level. But maybe adding arguments to the function call, for exemple imshow(x, set_contrast=True, linescan=True) plus keyboard toggles is enough? > >> agree to an extent. Nevertheless, I think providing a viewer that supplies >> *an infrastructure* for creating GUI tools might be really valuable. I >> haven't really thought through what this plugin system would look like, so >> it may just be a pipe dream. > This is a tricky problem (which set of tools to use, etc.)--but in > principle, it would be nice to have a few standard viewers, e.g., > image compare, image compare + parameter adjust, mask editor (to build > custom filters, e.g.) and image inspector (like the one you and > Guillaume are working on). > > Stéfan > What do you mean by mask editor? Something like a slider to adjust a threshold with a superimposed mask? Guillaume

5 4

Normalised cross correlation
by Malcolm Reynolds 06 Mar '12

06 Mar '12

Hi everyone, I've just stumbled across skimage as I'm going through to arduous process of trying to switch my workflow (PhD study of computer vision & machine learning) from matlab to python. The project looks great but I'm currently in need of a fast Normalised Cross Correlation function. As it is, I can't find any python implementation which is anywhere near the speed of Matlab's normxcorr2 (correct me if I'm wrong...) and it looks like a corresponding function to match the normxcorr2 behaviour is planned but not yet implemented for skimage. Is this correct? I'd be interested in looking into this and contributing some code to the project, but before I put a heap of time into it I just wanted to check that something didn't already exist in this or another numpy-based project. Cheers, Malcolm

2 1

Shining Panda continuous integration
by Stéfan van der Walt 05 Mar '12

05 Mar '12

Hi all, Have a look at our new continuous integration display at Shining Panda: https://jenkins.shiningpanda.com/skimage/ We're not doing too badly on the test coverage! https://jenkins.shiningpanda.com/skimage/job/skimage-linux-py32/44/cobertur… Stéfan

1 0

Re: harris example for 0.6dev missing a harris.
by Michael Aye 05 Mar '12

05 Mar '12

On Monday, March 5, 2012 3:49:41 PM UTC+1, Tony S Yu wrote: > > > > >> On Monday, March 5, 2012 11:37:16 AM UTC+1, Nelle wrote: >>> >>> Hi Michael, >>> >>> The example imports the harris method and not the module. It is a bit >>> confusing because the method and the module have both the same name (maybe >>> we can modify that ?). >>> >>> I've checked both on the documentation and on my machine, and the >>> example works fine. >>> What error message do you get when trying to run the example ? >>> >>> >> The documentation of the harris corner example on >> http://scikits-image.org/docs/dev/auto_examples/plot_harris.html >> states for the import: >> >> from skimage.feature import harris >> >> So when I do that, I get a module, with the method inside: >> >> In [149]: from skimage.feature import harris >> >> In [150]: harris >> Out[150]: <module 'skimage.feature.harris' from >> '/Users/maye/Library/Python/2.7/site-packages/skimage-0.4-py2.7-macosx-10.5-i386.egg/skimage/feature/harris.pyc'> >> > > Hi Michael, > > It appears that you're importing skimage 0.4 here. You may need to clean > out this older version and try again. > > -Tony > > >> >> Thanks, I had a messed up easy-install.pth. I wrongly assumed during a hot edit of that file that anything that starts with skimage is newer than something that starts with scikits_image and left the old entry up front newer later entries. Sorry for the noise. Michael > In [151]: type(harris) >> Out[151]: module >> >> In [152]: harris. >> harris.harris harris.ndimage harris.peak >> >> And the error is accordingly, a TypeError for the module not being >> callable: >> >> In [152]: harris(ctx.data) >> >> --------------------------------------------------------------------------- >> TypeError Traceback (most recent call >> last) >> /Users/maye/Dropbox/src/pymars/<ipython-input-152-a1e24f9ed74a> in >> <module>() >> ----> 1 harris(ctx.data) >> >> TypeError: 'module' object is not callable >> >> >> Regards, >> Michael >> >> >> Cheers, >>> N >>> >>> >>> >>> The example does an import of the harris module, not the class. >>>> Therefore for me in the code I have to call it like harris.harris(img, >>>> min_distance=6) >>>> >>>> Do I have a messed up system or is that an error? If yes, I can issue a >>>> PR. >>>> >>>> Michael >>>> >>>> >>>> >>> >

1 0

Re: harris example for 0.6dev missing a harris.
by Michael Aye 05 Mar '12

05 Mar '12

On Monday, March 5, 2012 11:37:16 AM UTC+1, Nelle wrote: > > Hi Michael, > > The example imports the harris method and not the module. It is a bit > confusing because the method and the module have both the same name (maybe > we can modify that ?). > > I've checked both on the documentation and on my machine, and the example > works fine. > What error message do you get when trying to run the example ? > > The documentation of the harris corner example on http://scikits-image.org/docs/dev/auto_examples/plot_harris.html states for the import: from skimage.feature import harris So when I do that, I get a module, with the method inside: In [149]: from skimage.feature import harris In [150]: harris Out[150]: <module 'skimage.feature.harris' from '/Users/maye/Library/Python/2.7/site-packages/skimage-0.4-py2.7-macosx-10.5-i386.egg/skimage/feature/harris.pyc'> In [151]: type(harris) Out[151]: module In [152]: harris. harris.harris harris.ndimage harris.peak And the error is accordingly, a TypeError for the module not being callable: In [152]: harris(ctx.data) --------------------------------------------------------------------------- TypeError Traceback (most recent call last) /Users/maye/Dropbox/src/pymars/<ipython-input-152-a1e24f9ed74a> in <module>() ----> 1 harris(ctx.data) TypeError: 'module' object is not callable Regards, Michael Cheers, > N > > > > The example does an import of the harris module, not the class. >> Therefore for me in the code I have to call it like harris.harris(img, >> min_distance=6) >> >> Do I have a messed up system or is that an error? If yes, I can issue a >> PR. >> >> Michael >> >> >> >

2 1

harris example for 0.6dev missing a harris.
by Michael Aye 05 Mar '12

05 Mar '12

The example does an import of the harris module, not the class. Therefore for me in the code I have to call it like harris.harris(img, min_distance=6) Do I have a messed up system or is that an error? If yes, I can issue a PR. Michael

2 1

Re: [scikits-image] Restore previous io plugins (#152)
by Stéfan van der Walt 02 Mar '12

02 Mar '12

On Fri, Mar 2, 2012 at 10:26 AM, Tony Yu <tsyu80(a)gmail.com> wrote: > Hmm, another issue: Using ``imshow(..., plugin='qt')`` doesn't behave > correctly (even after the PR above) because the ``show()`` function still > uses the default plugin (i.e. not necessarily the qt plugin), so the window > opened by ``imshow`` just disappears. We could stick a use_plugin call inside of qt's imshow. It's kind of nasty to adjust global state this way, but should fix the problem in the majority of cases. Stéfan

1 0