Commit Graph

521 Commits

Author SHA1 Message Date
David Hirvonen
cab0ecdf19 Avoid clearing HaarEvaluator::tofs variable (introduced fb8478eb7b) in HaarEvaluator::setImage(...) since it won't be recomputed by HaarEvaluator::computeOptFeatures() for cases where the input image dimensions and requested search scales are the same as the previous. If none of the above change then it should be correct, and if they do change it will be computed correctly inside HaarEvaluator::computeOptFeatures() (in commit 952f9dbe62)
I'm guessing there is a unit test for this that needs to be called 2x with the same image and search parameters (or another image with the same size) to illustrate the problem.
2015-05-13 17:34:16 -04:00
Michaël Sdika
ab40c2acf7 added the signedGradient varaible to the HOGDescriptor class.
when true, use signed gradient for the angular histogram computation.
default to false for backward compatibility.
2015-05-10 09:23:30 +02:00
Dikay900
952f9dbe62 initialize tiltedOffset variable at correct location
do not use this variable if it is not an tilted feature
2015-04-01 20:32:30 +02:00
Vadim Pisarevsky
3c51fdffac Merge pull request #3849 from Dikay900:fix_tilted_features_crash 2015-03-24 15:19:13 +00:00
Yan Wang
433c3da7cf Optimize the performance of cascade OpenCL kernel.
1. Use built-in mad() instead of += and *.
2. For stump stages, if weight.z == 0, avoid unnecessary calculation
because some features only have 2 rectangles..

It could improve
OCL_Cascade_Image_MinSize_CascadeClassifier.CascadeClassifier/*
about 10% mean values.

Signed-off-by: Yan Wang <yan.wang@linux.intel.com>
2015-03-23 17:47:40 +08:00
Dikay900
fb8478eb7b add resetting the offset variable after setting a new image 2015-03-20 13:26:13 +01:00
Dikay900
80069f384a resolve crash of haarclassifier when using a tilted feature 2015-03-19 20:47:33 +01:00
Naer Chang
cdc9bc397d Revert "issue 3710"
Trying to fix whitespace issue.
This reverts commit dcc2512f58.
2015-03-14 18:32:46 -04:00
Naer Chang
dcc2512f58 issue 3710
added a check to prevent illegal access to operator[] when vector is empty
2015-03-14 02:56:29 -04:00
Maksim Shabunin
37c74e38f4 Python support 2015-03-02 18:13:00 +03:00
Maksim Shabunin
457123027e Modified java wrapping mechanism 2015-03-02 18:12:30 +03:00
Vladislav Vinogradov
cda6fed41f move tegra namespace out of cv to prevent conflicts 2015-02-27 12:52:11 +03:00
Vladislav Vinogradov
44e41baffe use new functions before all tegra:: calls 2015-02-26 19:34:58 +03:00
Vadim Pisarevsky
d1a15097fe Merge pull request #3664 from wangyan42164:ocl_cascasde_barrier 2015-02-16 15:42:26 +00:00
Vadim Pisarevsky
6c8681a472 Merge pull request #3656 from SeninAndrew:master 2015-02-06 11:49:27 +00:00
Vadim Pisarevsky
1e47c6cc56 Merge pull request #3612 from guoyejun:fix_face_detect 2015-02-05 07:50:13 +00:00
Yan Wang
99aaf4030a Remove unnecessary barrier in runHaarClassifier().
Signed-off-by: Yan Wang <yan.wang@linux.intel.com>
2015-02-02 16:55:11 +08:00
Andrew Senin
52535b356c Fix for assert in compute when subimage is used (see issue #4149) 2015-01-29 21:39:07 +04:00
Yan Wang
df697f6520 Optimize runLBPClassifierStumpSimple by built-in mad24.
Signed-off-by: Yan Wang <yan.wang@linux.intel.com>
2015-01-22 16:36:16 +08:00
Vadim Pisarevsky
b49fa7d384 Merge pull request #3625 from wangyan42164:ocl_cascade_deadlock 2015-01-21 10:43:50 +00:00
Yan Wang
f5b21e8141 Avoid deadlock becasue some work item couldn't reach barrier in loop
and unsynchronized.

Signed-off-by: Yan Wang <yan.wang@linux.intel.com>
2015-01-20 13:45:33 +08:00
Vadim Pisarevsky
b6023eab9c Merge pull request #3594 from asmorkalov:opencv4tegra_support 2015-01-19 10:47:49 +00:00
Guo Yejun
595ec6b852 correct the size of buf which holds integral sum/sumsq
layer_dy is not initialized correctly, it results in a small height
of the buffer, and so memory write is out of range in cv::integral.
2015-01-16 10:51:39 +08:00
Yan Wang
07ab0c1c53 Avoid compile unnecessary OpenCL kernel.
Signed-off-by: Yan Wang <yan.wang@linux.intel.com>
2015-01-14 19:52:41 +08:00
Vadim Pisarevsky
f566f6baa9 Merge pull request #3535 from wangyan42164:use_hypot 2015-01-13 11:22:21 +00:00
Alexander Smorkalov
1fa37fe733 OpenCV for Tegra compilation fix for
- Gaussian blur;
- CascadeClassifier;
- Blenders in stitching module;
- Laplacial pyromids in stitching module.
2015-01-12 23:43:26 +03:00
Yan Wang
9699fde8d2 Use hypot to do vector calculation.
Signed-off-by: Yan Wang <yan.wang@linux.intel.com>
2014-12-30 14:12:28 +08:00
Maksim Shabunin
d01bedbc61 Removed Sphinx documentation files 2014-12-26 19:12:19 +03:00
Vadim Pisarevsky
d9f159a554 Merge pull request #3513 from mshabunin:compat-30 2014-12-22 11:58:01 +00:00
Maksim Shabunin
c485aee464 Included c-headers for better 2.4 compatibility 2014-12-19 17:05:26 +03:00
Yan Wang
efa84d8225 Use preprocessor for constant values in OpenCL kernel instead of
the parameter variable.

It could improve the performance of
OCL_Cascade_Image_MinSize_CascadeClassifier.CascadeClassifier/*.
Especially,
OCL_Cascade_Image_MinSize_CascadeClassifier.CascadeClassifier/15
OCL_Cascade_Image_MinSize_CascadeClassifier.CascadeClassifier/16
could be improved about 2% in Intel platform.

Signed-off-by: Yan Wang <yan.wang@linux.intel.com>
2014-12-16 16:21:05 +08:00
Maksim Shabunin
03e213ccae Doxygen documentation: BiB references and fixes 2014-12-01 15:50:55 +03:00
Maksim Shabunin
1523fdcc1c Doxygen documentation: more fixes and cleanups 2014-12-01 15:50:36 +03:00
Maksim Shabunin
8e9ea0e3d1 Doxygen doucmentation: ml module 2014-12-01 15:47:13 +03:00
Ari Suwendi
b14f7c99d2 bug fix for a crash when input image is small 2014-11-29 01:57:15 -08:00
Maksim Shabunin
ef3d02214b Fixing iOS clang warnings, part 2 2014-10-17 18:14:54 +04:00
Pavel Vlasov
45958eaabc Implementation detector and selector for IPP and OpenCL;
IPP can be switched on and off on runtime;

Optional implementation collector was added (switched off by default in CMake). Gathers data of implementation used in functions and report this info through performance TS;

TS modifications for implementations control;
2014-10-15 14:24:41 +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
Alexander Alekhin
55188fe991 world fix 2014-08-05 20:12:35 +04:00
Vadim Pisarevsky
2fd01dcced fixed doc builder warnings 2014-07-28 17:18:28 +04:00
Vadim Pisarevsky
1ad9827fc4 removed ERFilter (to be moved to opencv_contrib/modules/text) and lineMOD (to be moved to opencv_contrib/modules/rgbd) 2014-07-28 16:48:53 +04:00
vbystricky
4286f60387 Extract imgcodecs module from highgui 2014-07-07 16:28:08 +04:00
Vadim Pisarevsky
c3329ff79d Merge pull request #2856 from sanchom:master 2014-07-02 14:22:56 +00:00
Vadim Pisarevsky
c4fdf71bb3 fixed some more compile warnings on Windows 2014-06-25 20:22:56 +04:00
Vadim Pisarevsky
8f2aaff6d7 fixed compile warnings on Windows 2014-06-25 16:17:52 +04:00
Vadim Pisarevsky
66f2e527b6 removed extra whitespaces 2014-06-24 23:05:24 +04:00
Vadim Pisarevsky
07c0d50edf put the detection-based tracker back (to objdetect module), so that the Android face detection sample builds fine. The patch by Konstantin Matskevich that adds C++11 threads support has been merged in too. Also, fixed compile error in the doc builder. 2014-06-24 21:16:09 +04:00
Vadim Pisarevsky
3858f2291d removed contrib, legacy and softcsscade modules; removed latentsvm and datamatrix detector from objdetect. removed haartraining and sft apps.
some of the stuff will be moved to opencv_contrib module.
in order to make this PR pass buildbot, please, comment off opencv_legacy, opencv_contrib and opencv_softcascade test runs.
2014-06-24 19:18:51 +04:00
Sancho McCann
e1e243588e Bugfix: Memory leak in deletion of er_stack nodes of ERFilter. 2014-06-12 11:02:06 -07:00
Roman Donchenko
67e9708621 Merge pull request #2773 from abidrahmank:pycascade 2014-05-29 14:51:45 +04:00