[GRASS-user] i.superpixels.slic not working

[Moritz Lennert]

On 12/06/17 13:09, Moritz Lennert wrote:
> On 12/06/17 12:11, James Duffy wrote:
>>
>>
>> On 12 June 2017 at 11:00, Moritz Lennert <mlennert at club.worldonline.be
>> <mailto:mlennert at club.worldonline.be>> wrote:
>>
>>     On 12/06/17 11:38, James Duffy wrote:
>>
>>         On 12 June 2017 at 10:36, Moritz Lennert
>>         <mlennert at club.worldonline.be <mailto:mlennert at club.worldonline.be>
>>         <mailto:mlennert at club.worldonline.be
>>         <mailto:mlennert at club.worldonline.be>>> wrote:
>>
>>             On 12/06/17 11:25, James Duffy wrote:
>>
>>                 Hello,
>>
>>                 I'm trying to run the following:
>>
>>                 *i.superpixels.slic -n group=mod1 at PERMANENT output=test
>>                 num_pixels=5000
>>                 memory=10000
>>
>>                 *
>>                 The output gives me the following:
>>
>>                 (Mon Jun 12 10:17:40
>>                 2017)
>>                 i.superpixels.slic -n group=mod1 at PERMANENT output=test
>>                 num_pixels=5000
>>                 memory=10000
>>                 Cache data in memory mode
>>                 Loading input...
>>                 WARNING: Initialized 5036 of 5070 seeds
>>                 Performing k mean segmentation...
>>
>>                 And then I get a 'GRASS 7 has stopped working' box pop up.
>>
>>                 I tried again with the default settings for the module:
>>
>>                 *i.superpixels.slic group=mod1 at PERMANENT output=test
>>
>>                 *
>>                 And got the same error.
>>
>>                 I am using windows 10 professional and GRASS 7.3.svn (2017).
>>
>>                 Has anyone else got this module working, or had the same
>>         problem
>>                 on Windows?
>>
>>
>>             I haven't had any problems running it on Linux, and some
>>         colleagues
>>             have run it on Windows.
>>
>>             Just for our info: What are your region settings ?
>>
>>
>>         projection: 1 (UTM)
>>         zone:       43
>>         datum:      wgs84
>>         ellipsoid:  wgs84
>>         north:      25183.22184529
>>         south:      25157.71783908
>>         west:       301463.66480893
>>         east:       301494.2431551
>>         nsres:      0.07775612
>>         ewres:      0.0778075
>>         rows:       328
>>         cols:       393
>>         cells:      128904
>>
>>         Just to add, I am running i.segment.uspo without problems using
>>         the same
>>         imagery group and the same region.
>>
>>
>>     Yes, the region is actually very small, so no problem from there. It
>>     might be an issue with the memory handling. Could you try without
>>     setting the memory parameter, i.e. using its default value of 300
>>     and tell us what happens ?
>>
>>
>> *#default memory*
>> *i.superpixels.slic -n group=mod1 at PERMANENT output=test
>>
>> *
>> Produced the same problem.
>
> Can you reproduce using the NC demo data set ?
>
> I.e.
>
> g.region rast=ortho_2001_t792_1m
> i.group ortho input=ortho_2001_t792_1m
> i.superpixels.slic group=ortho out=slic_test
>
> And tying with and without the '-n' flag and the memory parameter.

Ok, I think I found the culprit: 
https://trac.osgeo.org/grass/changeset/70961/.

@Markus: something in that modification creates a segfault in memory 
mode when running the above commands. When I try with the version just 
before, it works.

Moritz