Open in new window / Try shogun cloud
--- Log opened Mon Nov 14 00:00:02 2011
-!- blackburn [~blackburn@] has quit [Quit: Leaving.]00:44
CIA-37shogun: Soeren Sonnenburg master * r0ef3680 / (7 files in 7 dirs): use free_matrix instead of destroy_matrix and adjust swig in typemaps -
shogun-buildbotbuild #348 of cmdline_static is complete: Success [build successful]  Build details are at
shogun-buildbotbuild #332 of octave_static is complete: Success [build successful]  Build details are at
shogun-buildbotbuild #332 of r_static is complete: Success [build successful]  Build details are at
shogun-buildbotbuild #324 of python_static is complete: Success [build successful]  Build details are at
-!- blackburn [5bdfb203@gateway/web/freenode/ip.] has joined #shogun09:13
blackburnsonney2k: sonne|work !09:13
blackburnis it working now?09:13
sonne|worklooks like...09:13
blackburnno leaks?09:14
sonne|workno idea if the memory leak is back09:14
blackburnsonne|work: can you please try valgrind --leak-check=full kernel_custom_modular or so?09:14
-!- blackburn is now known as blackburn|work09:14
sonne|workplease do ...09:14
sonne|workI don't have time for that right now09:14
blackburn|workI have no shogun available09:14
sonne|workthen grab it :)09:15
blackburn|workI'll have to install cygwin, etc09:15
sonne|workyou know it is open source ;-)09:15
sonne|workyeah - but it will work :)09:15
blackburn|worksonne|work: I have too bad connection for downloading whole cygwin and shogun :)09:16
blackburn|worksonne|work: okay I'll test it this night09:16
blackburn|workbut it would be bad if it was leaking again09:16
sonne|workbetter than crashing at least09:17
sonne|workbut my guess is that it isn't leaking09:17
blackburn|worksonne|work: leaking = crashing for me :)09:18
sonne|workwell YMMV09:40
sonne|workblackburn|work: did you ever do object detection in images?09:59
blackburn|worksonne|work: I'm currently starting to participate in project where we will do that10:00
blackburn|worksonne|work: but I have looked through literature already10:00
sonne|workso you know HOG features and LBP ?10:00
blackburn|worksonne|work: all I know is they exist :D10:01
blackburn|worksonne|work: are you doing ~ thing?10:02
sonne|workI think this would be a nice application10:02
sonne|workall these features are in opencv I guess - so easy to extract10:02
blackburn|workah, for shogun?10:02
sonne|workI mean say I would want to track an object10:02
sonne|workor detect traffic signs...10:03
blackburn|workI thought you are doing it on your job10:03
blackburn|workI guess you can't tell me what you do on job haha10:03
blackburn|worksonne|work: yes, I will work on traffic signs detection + recognition10:03
blackburn|worksonne|work: binding opencv and shogun is quite interesting, I agree10:05
sonne|workI mean we just use their features for images ... should be easy10:06
sonne|workproblem is all the insight about CV10:06
sonne|worknew field to me10:06
blackburn|workfor me too10:06
blackburn|worksonne|work: btw we (project group) want to find a way to not use HOG due to its time complexity10:07
blackburn|workit would be too slow for real-time traffic signs10:07
sonne|workdoesn't sound so slow10:09
blackburn|worksonne|work: right but no way to calculate HOG of the whole image.. same thing with SIFT10:12
blackburn|workI do not know much yet10:12
-!- supernaught [] has quit [Ping timeout: 248 seconds]10:12
blackburn|workI guess the only way is to try almost everything :)10:12
sonne|workblackburn|work: I really think computing HOG should be do-able...10:34
-!- supernaught [] has joined #shogun10:38
blackburn|worksonne|work: hmm okay I'll look through it more carefully10:52
blackburn|worksonne|work: so are you doing it at job? ;)10:52
-!- blackburn|work [5bdfb203@gateway/web/freenode/ip.] has quit [Quit: Page closed]11:13
-!- blackburn [5bdfb203@gateway/web/freenode/ip.] has joined #shogun12:19
blackburnsonne|work: how difficult it would be to make shogun compile-able under windozze without cygwin?12:20
blackburnsonne|work: I can't make shogun on cygwin due to no version_string.h file.. how to avoid this?12:44
blackburnwhy is it working with linux?12:44
blackburnwhoaa overwhelming warnings13:02
sonne|workno idea13:14
sonne|workbut go to shogun/13:14
sonne|workthen make lib/versionstring.h13:14
blackburnsonne|work: yeah did already13:18
blackburnbut CombinedBlaBlaBla kernel failed to compile13:18
sonne|worknot here13:18
sonne|workwhat is the error BLABLA?13:18
blackburnokay I'll copy13:19
blackburna lto of errors13:19
blackburnsonne|work: what 'not here'?13:19
blackburnc++: unrecognized option `-pthread' kernel/CommUlongStringKernel.cpp:1: warning: -fPIC ignored for target (all code is position independent) In file included from kernel/CommUlongStringKernel.cpp:11: ../shogun/lib/common.h:48:8: warning: undefining "__STDC_FORMAT_MACROS" kernel/CommUlongStringKernel.cpp:19: error: `CCommUlongStringKernel' has not bee n declared kernel/CommUlongStringKernel.cpp:20: error: ISO C++ forbids declaration 13:21
sonne|workdon't you have shogun/kernel/CommUlongStringKernel.h   ?13:24
blackburnsonne|work: I have13:24
sonne|workbut how come it doesn't know about this class then?13:24
blackburnsonne|work: no idea..13:26
sonne|workand besides it was working on my windows box13:27
blackburnsonne|work: I'll make a try on my notebook's win713:40
blackburnsonne|work: got it13:43
blackburnsonne|work: uint64_t is disabled13:43
sonne|workerr and?13:43
blackburnsonne|work: CommULongblabla is derived from StringFeatures<uint64_t>13:44
blackburnah no13:44
blackburnI got wrong13:44
blackburnuint64_t is enabled13:44
blackburnthen no idea what is wrong13:44
sonne|workthese uint* enabled/disabled stuff should only affect the interfaces13:45
sonne|worknot shogun13:45
blackburnremoved it temporary13:59
sonne|workwhich version of g++ is this?14:00
blackburnThread model: posix gcc version 3.4.4 (cygming special, gdc 0.12, using dmd 0.125)14:01
sonne|workok then14:01
sonne|workuse gcc 4.x!14:01
sonne|workI am pretty sure gcc 3.x is #*($$UY(!14:01
blackburnoh okay14:01
sonne|workI am a bit surprised - I thought gcc4* is in latest cygwin for a few years? how come you got gcc3?14:13
blackburnsonne|work: checked in gcc14:14
sonne|workblackburn: btw
sonne|workHOG features in real time...14:32
blackburnsonne|work: sure but CUDA on quadro14:32
sonne|workyeah sure14:33
sonne|workI guess you can easily do the usual cascade/prefiltering14:34
blackburnI know that there are israeli embedded device with no cuda, etc14:34
blackburnmaking lane detection, etc14:34
blackburnoh I really want to test HOG on images14:38
sonne|worksome other intersting features are ORB features14:46
sonne|worksimilar to SIFT but patent free14:47
sonne|workand LBP14:47
sonne|workI guess these should be powerful enough to do any kind of detection14:47
sonne|workthe last slide looks impressive (the avg gradient)
blackburnsonne|work: nice15:00
sonne|workkind of an alien :)15:00
blackburnI hope it would work for road signs as well15:06
-!- alesis-novik [] has joined #shogun15:41
sonne|workblackburn: I mean one could at least try15:45
blackburnI should try15:46
sonne|worklooks like opencv even has some python bindings for these features...15:46
blackburn 16:05
blackburn???, ?? ???? :)16:07
blackburnsonne|work: ? ????????????? shogun16:07
sonne|workI like that word :) kompilirowatch :)16:09
blackburnsonne|work: kompilirowat'16:14
sonne|worknet ne mjankisak?16:15
blackburn????????? was ist das?16:15
blackburn?????? ????? :)16:16
blackburn????????????? - ????????????? if I catch you right16:17
blackburnkompilirowat' - skompilirowaL16:17
blackburnsonne|work: ich spreche deutsch etwas! :)16:18
sonne|workja perfekt :)16:19
blackburnsind sie mude? ich bin mude!16:19
blackburner du mude?16:20
blackburnsie = you in plural, right?16:20
blackburnSie = You in singular16:20
sonne|workwell sie16:20
blackburnsonne|work: is "er du"  correct?16:20
sonne|worksie == du16:20
sonne|workbist du muede would be the right thing16:21
blackburnI see16:23
blackburnI always laugh on 'gluklich' word16:23
blackburnsonne|work: ???? in russian slang is some gallucination :D16:23
blackburnsonne|work: bist du glueklich?16:34
blackburntime to leave job!16:40
-!- blackburn [5bdfb203@gateway/web/freenode/ip.] has quit [Quit: Page closed]16:58
-!- in3xes [~in3xes@] has joined #shogun19:15
-!- blackburn [~blackburn@] has joined #shogun19:43
@sonney2kblackburn, so whats up with the leak(tm)19:50
@sonney2kall good now?19:50
blackburnsonney2k: I'll check soon19:50
blackburnjust came home19:50
@sonney2kI am a bit worried about the clang == crash issue19:50
blackburnhuh 10-50 and I came home19:51
@sonney2klooks like things are not crashing under debian sid :)20:01
blackburnsonney2k: I'll check for leak in a min20:02
blackburnsonney2k: I've got access to color FERET database20:02
@sonney2kis this public?20:03
blackburnsonney2k: semi-public I would say - you need to require access20:04
blackburnsonney2k: python went smoothly20:05
blackburnsonney2k: leak(tm) is gone20:09
@sonney2kclang's warnings found another round of bugs...20:10
blackburnI'm now off cause I'm deadly hungry but later we would fix some20:14
CIA-37shogun: Soeren Sonnenburg master * r6c75752 / src/interfaces/ruby_modular/swig_typemaps.i : use strcmp for string comparison in ruby typemaps -
CIA-37shogun: Soeren Sonnenburg master * r56afbe6 / (2 files in 2 dirs): enable set_shifts in wds examples -
blackburnsonney2k: ready for bugfixing21:19
blackburnsonney2k: :(21:30
shogun-buildbotbuild #47 of nightly_default is complete: Success [build successful]  Build details are at
@sonney2kblackburn, feel free ... I noticed that sth. is wrong with lua - crashes / memory hog ...21:59
blackburnI don't know what are the bugs21:59
blackburnlua crashes?21:59
blackburnwas it working?21:59
@sonney2krun make check-examples and you will see - then we have the lib/versionstring.h issue22:00
@sonney2kthen some old tests fail22:00
@sonney2kblackburn, do you remember if we fixed all tests before we release 1.0.0?22:01
blackburnsonney2k: which tests?22:01
@sonney2kahh btw you need to do your examples / tests for dimred methods...22:01
@sonney2kthe tests you get when you run make tests22:02
blackburnsonney2k: they compare with tests from data, right?22:06
@sonney2ktests are based on data if you meant that22:08
blackburnsonney2k: yes22:09
blackburnsonney2k: will you have a few minutes on next week to review my paper on KLTSA?22:09
-!- alesis-novik [] has quit [Ping timeout: 240 seconds]22:27
-!- alesis-novik [] has joined #shogun22:38
-!- in3xes [~in3xes@] has quit [Ping timeout: 260 seconds]22:48
-!- in3xes [~in3xes@] has joined #shogun22:49
-!- alesis-novik [] has quit [Quit: Leaving]23:10
@sonney2kI will try my best... but I am not an expert on this subject I am afraid23:11
blackburnsonney2k: there is nothing very deep here, just simple trick23:24
blackburnI will try to publish it on arxiv23:25
blackburnand then cite in paper with chris23:25
@sonney2kwell arxiv is not a journal or anything just some repository23:26
@sonney2kso it is not counted as a paper... but techreport23:27
blackburnsonney2k: yes23:27
blackburnsonney2k: but recall I'm newbie and my tech report will look like a shit w/o reviewing23:28
blackburnsonney2k: so let it call tech report but anyway I would appreciate :)23:28
blackburnsonney2k: I have no time/skillz to make truly paper and I don't think this idea worths a paper23:31
@sonney2kblackburn, go fix your tests23:32
@sonney2kotherwise we will never be able to release 1.123:32
blackburnsonney2k: ?? ????23:39
blackburnsonney2k: it is you23:39
@sonney2kI am more like
@sonney2kanyways time to sleep23:41
blackburnyou are23:41
blackburnyou get me frustrated and not believing I could do anything!23:42
blackburnI would better go sleep23:42
-!- blackburn [~blackburn@] has quit [Quit: Leaving.]23:42
--- Log closed Tue Nov 15 00:00:07 2011