[IPOL discuss] MATLAB demos

Jacques Froment Jacques.Froment at univ-ubs.fr
Tue Jun 2 13:40:31 CEST 2015 

Dear Luis,

I thank you for these considerations I was not aware of.

I completely agree with you about the fact that submitting a paper to IPOL is a [too] complex task.
But in my opinion the main reason is the need for authors to submit a demo and thus to have to write
code in Python and html that has nothing to do with the submitted algorithm. I believe that this task
should not be left to the authors.

Going back to the question of Octave compatibility, writing this compatibility as a recommendation
would be better than nothing.

Best,
Jacques.

Jacques Froment   (Jacques.Froment at univ-ubs.fr)

Université de Bretagne Sud - LMBA UMR 6205
Centre Y. Coppens, campus de Tohannic
BP 573, F-56017 VANNES, FRANCE
Tél. 0297017138 (+33 297017138)
Fax 0297017175 (+33 297017175)

On 02/06/2015 11:00, Luis Alvarez wrote:
> Dear all,
>
> The main challenge we face at the current moment is to increase the number of papers and the number of new authors who
> send papers to IPOL. The current IPOL paper publication statistic is :
>
> 2011 : 21 papers
> 2012: 15 papers
> 2013: 16 papers
> 2014: 21 papers
> 2015: 3 papers (up to now)
>
> To improve these figures is mandatory if we want to consolidate IPOL as an international journal and to submit IPOL to
> the JCR and SCOPUS Journal database evaluation. Otherwise, in my opinion, there is a high risk that IPOL becomes
> irrelevant or even disappears. In this regards I think that to accept new languages as Matlab (which is used by a large
> community) could be an important step in this direction. Of course it would be nice that the Matlab code be compatible
> with Octave but I think that such compatibility should not be mandatory, it should be written as a recommendation.
> Please do not forget that currently, to send a paper to IPOL for somebody who is not familiar with the "IPOL Community"
> is a quite complex task. To encourage new authors to send papers to IPOL we have to ease the paper edition procedure and
> to clearly differentiate recommendations from mandatory conditions which should be reduced as much as possible.
>
> Regards
> Luis Alvarez
>
>> Dear Miguel,
>>
>> Exactly, my suggestion is basically to allow only Matlab toolboxes that
>> have the same signature and behavior as those available on Octave-Forge.
>> More precisely, since all functions of a given Matlab toolbox may not
>> be implemented in the corresponding Octave-Forge package, I mean to
>> allow only functions implemented in Octave-Forge that give the same
>> results than the ones in Matlab toolboxes.
>>
>> I believe that this compatibility may be checked almost automatically at
>> the editing step by checking that the source code gives the same outputs
>> (within a given numerical accuracy) running on Octave.
>>
>> I agree that this compatibility requirement would add a bit of work for
>> authors and editors, but it would be a "political" statement in favor of
>> free and open source softwares and it would not disadvantage labs that
>> do not have Matlab license (which is rather frequent outside the US and
>> elsewhere, outside major labs).
>>
>> Best,
>> Jacques.
>>
>> Jacques Froment   (Jacques.Froment at univ-ubs.fr)
>>
>> Université de Bretagne Sud - LMBA UMR 6205
>> Centre Y. Coppens, campus de Tohannic
>> BP 573, F-56017 VANNES, FRANCE
>> Tél. 0297017138 (+33 297017138)
>> Fax 0297017175 (+33 297017175)
>>
>> On 01/06/2015 20:19, Miguel Colom wrote:
>>> Quoting Jacques Froment <Jacques.Froment at univ-ubs.fr>:
>>>
>>>> Dear all,
>>>>
>>>> I am very reserved about the possibilities of publishing a code
>>>> that needs a commercial software. Would it be possible to request
>>>> the source code to be also compatible with the free GNU Octave
>>>> software ? (as Octave treats incompatibility with Matlab as a bug,
>>>> the main impact would be a limitation of usable functions from Matlab
>>>> toolboxes).
>>>
>>> Dear Jacques,
>>> We haven't finished writing the guidelines for MATLAB submissions and therefore we're open to any kind of suggestions.
>>>
>>> About forcing compatibility to Octave, I think that will depend on if the toolboxes that we'll allow have the same
>>> signature and behavior as in  Matlab. If not, then we can't ask it.
>>>
>>> But as I say, the rules haven't been already fixed.
>>>
>>> Best,
>>> Miguel
>>>
>>>
>>>>
>>>> Sincerely,
>>>> Jacques.
>>>>
>>>> Jacques Froment   (Jacques.Froment at univ-ubs.fr)
>>>>
>>>> Université de Bretagne Sud - LMBA UMR 6205
>>>> Centre Y. Coppens, campus de Tohannic
>>>> BP 573, F-56017 VANNES, FRANCE
>>>> Tél. 0297017138 (+33 297017138)
>>>> Fax 0297017175 (+33 297017175)
>>>>
>>>> On 01/06/2015 15:10, Miguel Colom wrote:
>>>>> Dear all,
>>>>> IPOL is accepting MATLAB demos now.
>>>>>
>>>>> We're working on writing clear guidelines for the MATLAB source codes (accepted toolboxes, style, etc) but meanwhile
>>>>> it's already possible to submit MATLAB codes and demos for publication.
>>>>>
>>>>> This message is only to explain what to do to have a demo executing MATLAB.
>>>>>
>>>>> The MATLAB code must be compiled. This step is exactly the same as what we do for any other compiled language and
>>>>> to do
>>>>> it it suffices to write a Makefile which calls mcc (the MATLAB compiler).
>>>>>
>>>>> For example:
>>>>> PROG=gauss
>>>>> all:
>>>>>     mcc -m ${PROG}.m -a lib -R -nodisplay
>>>>>
>>>>> The generated binary must be moved to the "bin" directory, as usual.
>>>>>
>>>>> The compiler generates not only the binary, but also a BASH script which calls the binary within the proper
>>>>> environment.
>>>>> You don't need to care about what the script does. Think of it as a black box where you put your parameters and it
>>>>> executes your program.
>>>>>
>>>>> The script needs the following parameters, in this order:
>>>>> 1) The MATLAB path
>>>>> 2) The parameters of your algorithm.
>>>>>
>>>>> The MATLAB path can be obtained from the demo system configuration, by calling self.get_MATLAB_path().
>>>>>
>>>>> For example:
>>>>> p = self.run_proc(['run_gauss.sh', self.get_MATLAB_path(), 'input_0.png',
>>>>>                    'output_1.png', str(sigma)])
>>>>>
>>>>> You need also to configure the path of MATLAB with two new entries in demo.conf.
>>>>> In the production server the configuration is:
>>>>>
>>>>> # demo paths
>>>>> demo.extra_path = '/usr/bin:/bin:/wheezy/usr/local/MATLAB'
>>>>> demo.matlab_path = '/wheezy/usr/local/MATLAB'
>>>>>
>>>>> You need to put there the right paths corresponding to your MATLAB installation.
>>>>>
>>>>> And you need also to have an updated version of the demo system.
>>>>> You can download a copy with: git clone https://github.com/mcolom/ipolDevel.git
>>>>> In principle you only need to change the contents of the "lib" directory.
>>>>>
>>>>> All the best,
>>>>> Miguel
>>>>>
>>>>>
>>>>> --
>>>>> IPOL - Image Processing On Line   - http://ipol.im/
>>>>>
>>>>> contact     edit at ipol.im          - http://www.ipol.im/meta/contact/
>>>>> news+feeds  twitter @IPOL_journal - http://www.ipol.im/meta/feeds/
>>>>> announces   announce at list.ipol.im - http://tools.ipol.im/mm/announce/
>>>>> discussions discuss at list.ipol.im  - http://tools.ipol.im/mm/discuss/
>>>> --
>>>> IPOL - Image Processing On Line   - http://ipol.im/
>>>>
>>>> contact     edit at ipol.im          - http://www.ipol.im/meta/contact/
>>>> news+feeds  twitter @IPOL_journal - http://www.ipol.im/meta/feeds/
>>>> announces   announce at list.ipol.im - http://tools.ipol.im/mm/announce/
>>>> discussions discuss at list.ipol.im  - http://tools.ipol.im/mm/discuss/
>>>>
>>>
>>>
>>>
>>> --
>>> IPOL - Image Processing On Line   - http://ipol.im/
>>>
>>> contact     edit at ipol.im          - http://www.ipol.im/meta/contact/
>>> news+feeds  twitter @IPOL_journal - http://www.ipol.im/meta/feeds/
>>> announces   announce at list.ipol.im - http://tools.ipol.im/mm/announce/
>>> discussions discuss at list.ipol.im  - http://tools.ipol.im/mm/discuss/
>> --
>> IPOL - Image Processing On Line   - http://ipol.im/
>>
>> contact     edit at ipol.im          - http://www.ipol.im/meta/contact/
>> news+feeds  twitter @IPOL_journal - http://www.ipol.im/meta/feeds/
>> announces   announce at list.ipol.im - http://tools.ipol.im/mm/announce/
>> discussions discuss at list.ipol.im  - http://tools.ipol.im/mm/discuss/
>
> --
> IPOL - Image Processing On Line   - http://ipol.im/
>
> contact     edit at ipol.im          - http://www.ipol.im/meta/contact/
> news+feeds  twitter @IPOL_journal - http://www.ipol.im/meta/feeds/
> announces   announce at list.ipol.im - http://tools.ipol.im/mm/announce/
> discussions discuss at list.ipol.im  - http://tools.ipol.im/mm/discuss/

More information about the discuss mailing list