Commit Graph

246 Commits

Author SHA1 Message Date
Vadim Pisarevsky
c9ccf3b55d Merge pull request #5346 from art-programmer:art-programmer-patch-1 2015-09-21 12:06:46 +00:00
art-programmer
e0ef293645 Update em.cpp
Fix a bug. When reading from a saved model, function decomposeCovs() will be called. And if covMatType is COV_MAT_DIAGONAL, covsEigenValues is computed using SVD and eigen values are sorted so that the order of eigen values is not preserved. This would lead to different result when calling function predict2. This issues is discussed here: http://stackoverflow.com/questions/23485982/got-different-empredict-results-after-emread-saved-model-in-opencv.
2015-09-14 19:35:53 -05:00
Ilya Lavrenov
ee2008abe8 fixed "Conditional jump or move depends on uninitialised value(s)" in GBD 2015-09-13 12:36:57 +02:00
Ilya Lavrenov
1d77d697ef fixed memory leak in GBTrees 2015-09-13 12:15:30 +02:00
Lorena García
c545857ea6 Indent fix 2015-08-04 16:50:55 +02:00
Lorena García
4cd78bc7ad Whitespace error 2015-08-04 16:45:39 +02:00
Lorena García
fe62243805 Whitespace error 2015-08-04 16:37:41 +02:00
Lorena García
2cb47b4191 Update data.cpp
Update to allow loading data from a .csv file that only contains training samples values and not the associated responses. To allow this behavior, function must be called with parameters ResponsesStartIdx <= -2 and ResponsesEndIdx = 0
2015-08-04 15:58:24 +02:00
Jie Yang
2ba0d31785 Fix SVM::trainAuto (Bug #4464)
Fix SVM::trainAuto (Bug #4464)
2015-07-06 01:20:26 +09:30
berak
b29e473c21 fix plausability check, #issue 4429 2015-07-01 15:32:37 +02:00
Maksim Shabunin
b95769333c Fixed ANN_MLP dw initialization when moment is not 0 2015-05-13 16:56:34 +03:00
Vadim Pisarevsky
a69b435c92 Merge pull request #3909 from Nuzhny007:master 2015-04-21 14:02:59 +00:00
Maksim Shabunin
cb00c7036f Fix cache error during SVM train 2015-04-15 15:46:07 +03:00
Сергей Нужный
4a65f3cee2 Squashed commit of the following:
commit 5c379aa1b8
Merge: e800800 3646967
Author: Сергей Нужный <snuzhny@stc-spb.ru>
Date:   Wed Apr 8 12:44:46 2015 +0300

    Merge branch 'master' of https://github.com/Nuzhny007/opencv

commit e800800bec
Author: Сергей Нужный <snuzhny@stc-spb.ru>
Date:   Wed Apr 8 12:43:38 2015 +0300

    Squashed commit of the following:

    commit 3646967deb
    Author: Сергей Нужный <snuzhny@stc-spb.ru>
    Date:   Tue Apr 7 15:48:30 2015 +0300

        Fix bug in SVM::trainAuto

    commit 0c24ccf538
    Author: Сергей Нужный <snuzhny@stc-spb.ru>
    Date:   Tue Apr 7 15:43:59 2015 +0300

        Revert "Fix bug in SVM::trainAuto"

        This reverts commit 225da0226e.

    commit 225da0226e
    Author: Сергей Нужный <snuzhny@stc-spb.ru>
    Date:   Tue Apr 7 15:09:47 2015 +0300

        Fix bug in SVM::trainAuto

commit 3646967deb
Author: Сергей Нужный <snuzhny@stc-spb.ru>
Date:   Tue Apr 7 15:48:30 2015 +0300

    Fix bug in SVM::trainAuto

commit 0c24ccf538
Author: Сергей Нужный <snuzhny@stc-spb.ru>
Date:   Tue Apr 7 15:43:59 2015 +0300

    Revert "Fix bug in SVM::trainAuto"

    This reverts commit 225da0226e.

commit 225da0226e
Author: Сергей Нужный <snuzhny@stc-spb.ru>
Date:   Tue Apr 7 15:09:47 2015 +0300

    Fix bug in SVM::trainAuto
2015-04-08 12:51:57 +03:00
Vadim Pisarevsky
052593c760 1. Input/OutputArray optimizations;
2. Algorithm::load/save added (moved from StatModel)
3. copyrights updated; added copyright/licensing info for ffmpeg
4. some warnings from Xcode 6.x are fixed
2015-04-07 16:44:26 +03:00
Sancho McCann
6593422c05 Bugfix: #4030 SVM auto-training. 2015-04-01 20:44:51 +00:00
Evgeny Agafonchikov
91bd7df8c1 Fixing invalid opencv_test_ml calls 2015-03-20 17:49:03 +03:00
ippei ito
518fd59a10 fixed whitespace error 2015-03-12 01:09:15 +09:00
ippei ito
49b38fe793 Fixed compile error build with VisualStudio2013 on windows.
REG_NONE defined in ml.hpp conflicts with defined in Winnt.h
2015-03-12 00:32:17 +09:00
Maksim Shabunin
79e8f0680c Updated ml module interfaces and documentation 2015-02-17 11:46:14 +03:00
Maksim Shabunin
c85197da55 Fixed allocated vector size in ML ANN algorithm
- updated return value of points_classifier sample
2015-02-03 18:16:20 +03:00
Maksim Shabunin
38d37d2b00 Fixed uninitialized memory usage in ANN_MLP 2014-12-31 12:16:26 +03:00
Maksim Shabunin
d004ee58c5 Support loading old models in ML module
- added test for loading legacy files
- added version to new written models
- fixed loading of several fields in some models
- added generation of new fields from old data
2014-12-31 12:16:25 +03:00
Julien Nabet
2d5e3b57ea Fix cppcheck reports warnings:
(performance) Possible inefficient checking for 'lmap' emptiness
(warning) Assignment of function parameter has no effect outside the function. Did you forget dereferencing it?
(style) Array index '<xxx>' is used before limits check
2014-12-27 10:54:34 +01:00
bartaandras
21227ba092 Fixed calc_output_scale with NO_OUTPUT_SCALE flag set. 2014-11-04 11:35:45 +01:00
Vadim Pisarevsky
b37a641cc9 fixed bug in automatic SVM training 2014-10-18 20:53:39 +04:00
Benoit Seguin
450734cc32 Correction of the integer data treated as floating point data bug, already corrected in the 2.4 branch. 2014-09-23 14:03:20 +02:00
Vadim Pisarevsky
4057e27539 Merge pull request #3126 from avdmitry:move_KDTree_to_ml 2014-09-14 18:57:23 +00:00
Dmitriy Anisimov
a6b29be55b minor change: moved algorithm type to the end of params 2014-09-13 15:06:07 +04:00
Dmitriy Anisimov
dfaf75f502 moving algorithm type to param 2014-09-06 09:29:32 +04:00
Dmitriy Anisimov
5f3ee657ce removed kdtree declaration from interface 2014-08-31 21:39:47 +04:00
Dmitriy Anisimov
9ddb23e025 first implementation KNearest wrapper on KDTree 2014-08-23 18:41:32 +04:00
Dmitriy Anisimov
37b1a7560c first version of moving KDTree from core to ml 2014-08-18 22:40:31 +04:00
Maksim Shabunin
108caae216 Modified logistic regression module according to comments
- Reworked documentation to reflect actual code
- Removed some unused variables
- Removed unnecessary 'cv::' modifiers
2014-08-18 19:13:59 +04:00
Maksim Shabunin
4667e18831 Updated logistic regression example
- Extracted common operations to separate functions.
- Activated first parameters set.
- Some output formatting.
- Fixed loop break condition in mini_batch_gradient function.
2014-08-18 19:06:58 +04:00
Maksim Shabunin
3e26086f82 Reworked ML logistic regression implementation, initial version 2014-08-18 19:06:58 +04:00
Maksim Shabunin
71770eb790 Fixed ML module build after merge 2014-08-18 19:06:58 +04:00
Rahul Kavi
a23836231f fixed warnings in type conversions
fixed warnings in type conversions from size_t to int (in getting size of number of unique classes in a training problem).
2014-08-18 19:06:58 +04:00
Rahul Kavi
d20b2a5a97 removed extra comments in train method 2014-08-18 19:06:58 +04:00
Rahul Kavi
ae02ecec60 removed a couple of unnecessary comments in Logistic Regression training method 2014-08-18 19:06:58 +04:00
Rahul Kavi
65eb52a247 removed default value from LogisticRegression constructor function definition 2014-08-18 19:06:57 +04:00
Rahul Kavi
8995921cc5 fixed white space in logistic regression classifier 2014-08-18 19:06:57 +04:00
Rahul Kavi
b3b4e83aed updated logistic regression definition 2014-08-18 19:06:57 +04:00
Rahul Kavi
5bb43f69ab removed trailing whitespaces from predict function 2014-08-18 19:06:52 +04:00
Rahul Kavi
0e13f33193 rewrote the code to update API, removed inheritance from CvStatModel. 2014-08-18 19:06:48 +04:00
Rahul Kavi
6c74439d56 fixed trailing white-space in logistic regression class 2014-08-18 19:06:48 +04:00
Rahul Kavi
6ae43a2243 added logistic regression classifier 2014-08-18 19:06:48 +04:00
Adil Ibragimov
8a4a1bb018 Several type of formal refactoring:
1. someMatrix.data -> someMatrix.prt()
2. someMatrix.data + someMatrix.step * lineIndex -> someMatrix.ptr( lineIndex )
3. (SomeType*) someMatrix.data -> someMatrix.ptr<SomeType>()
4. someMatrix.data -> !someMatrix.empty() ( or !someMatrix.data -> someMatrix.empty() ) in logical expressions
2014-08-13 15:21:35 +04:00
Vadim Pisarevsky
05e7c29de5 fixed various warnings and obvious errors reported by clang compiler and the coverity tool. 2014-08-10 00:10:05 +04:00
Vadim Pisarevsky
222f8a3d5e hopefully fixed test failure on Linux 2014-08-04 01:16:45 +04:00