Commit Graph

259 Commits

Author SHA1 Message Date
peng xiao
fd7ba355ee Add non-stump based ocl Haar cascade classifier support.
For example, haarcascade_frontalface_alt2.xml is now supported.
Note that classifier's pattern of a cascade file must be consistent,
i.e., all trees must either have two nodes or one node, otherwise
unexpected results will occur.

Other fixes:
Test cases are updated.
Some unused codes are removed.
Fix some problems of haar when using OclCascadeClassifierBuf.
2013-05-30 14:01:19 +08:00
peng xiao
006e4242b2 Merge branch '2.4' of https://github.com/Itseez/opencv into 2.4_oclgfft 2013-05-29 17:57:14 +08:00
peng xiao
d015fa76fa Fix 2.4 ocl Canny.
This fix is a workaround for current 2.4 branch without introducing an
additional oclMat buffer into CannyBuf object.
Test case is cleaned up.
Volatile keywords in kernels are removed for performance concern.
2013-05-29 14:15:26 +08:00
Vadim Pisarevsky
2ccdf56119 Merge pull request #913 from janm399:2.4 2013-05-28 22:05:14 +04:00
Vadim Pisarevsky
24ca620c32 Merge pull request #916 from bitwangyaoyao:2.4_fixPyrLK 2013-05-28 22:04:37 +04:00
yao
d81c145fa9 fix memory leak 2013-05-28 18:07:41 +08:00
peng xiao
1d0c283508 Fix a bug when pushing pointers of arguments into std::vector.
When argument pointers pushed into an vector and the pointers point to
address on stack, we need to make sure they are valid until kernels are
successfully flushed onto the queue.
2013-05-28 17:27:55 +08:00
peng xiao
6fae02c05d Fix some OpenCL kernel file build errors on Mac. 2013-05-28 11:12:05 +08:00
yao
c58e0d5d73 fix hog 2013-05-24 13:46:21 +08:00
yao
d45f9ef866 fix Linux build errors 2013-05-23 17:58:50 +08:00
yao
a223b5624f fix pyrLK's mismatch on Intel GPUs 2013-05-23 10:55:08 +08:00
Jan Machacek
46b770f255 Fixed include name in OpenCL on OS X 2013-05-22 13:22:16 +01:00
Roman Donchenko
519ee72f6f Merge pull request #905 from bitwangyaoyao:2.4_TVL1 2013-05-22 12:08:16 +04:00
peng xiao
b4a4a05bdc Add ocl's good features to track implementation.
Additional notes with this commit:
1. Add cornerHarris_dxdy and cornerMinEigenVal_dxdy to get
the interim dx and dy output of Sobel operator;
2. Add minMax_buf to allow user to reuse buffers in minMax;
3. Fix an error when either min or max pointer fed into minMax is NULL;
4. Corner sorter temporarily uses C++ STL's quick sort. A parallel
 selection sort in OpneCL is contained in the implementation but disabled
due to poor performance at the moment.
5. Accuracy test for ocl gfft.
2013-05-22 13:46:42 +08:00
Vadim Pisarevsky
cbbc82a789 Merge pull request #882 from pengx17:2.4_rewrite_query_info 2013-05-20 14:01:03 +04:00
yao
d8b192c84d Fix the mismatch on NV GPUs 2013-05-20 14:46:17 +08:00
yao
4162ebfad3 add OpticalFlowDual_TVL1_OCL function 2013-05-17 15:34:22 +08:00
peng xiao
1ecc765903 Merge branch '2.4' of https://github.com/Itseez/opencv into 2.4_pyrup_fix 2013-05-15 08:55:43 +08:00
peng xiao
df3997b108 Fix ocl::pyrUp
Use predefined OpenCL function to convert integers to floating points.
This is more accurate than before as it enables:
1. saturate cast
2. customized rounding
2013-05-15 08:51:21 +08:00
Vadim Pisarevsky
25e6902a43 Merge pull request #818 from bitwangyaoyao:2.4_optBlur 2013-05-14 15:56:56 +04:00
Vadim Pisarevsky
bd1d7cd208 Merge pull request #839 from pengx17:2.4_ocl_csbp 2013-05-14 15:56:07 +04:00
Vadim Pisarevsky
87765c0f16 Merge pull request #883 from bitwangyaoyao:2.4_fixMoments 2013-05-14 15:55:38 +04:00
yao
1d1d28baf2 fix black screen when input Mat is large 2013-05-14 18:02:53 +08:00
peng xiao
ac21cabda2 Copy ocl::queryDeviceInfo interface from master to 2.4.
Affected functions surf.ocl, pyrlk.ocl and hog.ocl are updated with the change.
2013-05-14 17:50:38 +08:00
Vadim Pisarevsky
d79c05fec9 Merge pull request #872 from pengx17:2.4_create2dimage_fix 2013-05-13 23:12:25 +04:00
Vadim Pisarevsky
fd83f2f5ca Merge pull request #819 from bitwangyaoyao:2.4_haarBuf 2013-05-13 22:36:10 +04:00
Andrey Kamaev
2665c39a0d Fix build warnings from gcc 4.8 2013-05-13 17:07:30 +04:00
Vadim Pisarevsky
9cce8ca4b6 Merge pull request #864 from pengx17:2.4_stereobm 2013-05-12 22:16:03 +04:00
peng xiao
e14acabb3d Fix ocl moudle build with latest Intel OpenCL SDK.
We found that cl.h file provided in latest Intel SDK muted deprecated interfaces.
2013-05-09 14:00:36 +08:00
peng xiao
69e6d0016e Optimize stereobm a bit.
Speedup about 30% on 6730M GPU.
2013-05-08 17:29:24 +08:00
peng xiao
d34e7eca60 Suppress warning when compiling deprecated OpenCL function on GNU compilers. 2013-05-06 17:16:45 +08:00
peng xiao
ed2199a497 Fix build 2013-05-03 09:54:11 +08:00
peng xiao
1eca49f40b ocl: Enable backward binary portability for setTo function. 2013-05-03 09:45:56 +08:00
peng xiao
168c0b0385 Optimize ocl::stereobm.
1. Use macro defines for some parameters(radius).
2. Reduce local memory usage.
3. Fix accuracy problem on Intel GPU.
2013-05-02 16:14:28 +08:00
peng xiao
355bc691fc Add OpenCL version 1.2 query into ocl::Context::supportsFeature().
Add backwards portability for OpenCL 1.1 when OpenCV executables
are compiled with OpenCL 1.2 profile support.
2013-05-02 14:44:59 +08:00
Vadim Pisarevsky
2068c4582c Merge pull request #844 from bitwangyaoyao:2.4_integral 2013-04-28 00:37:40 +04:00
yao
f788d010f7 fix a crash on CPU OCL for some specified image sizes 2013-04-24 14:12:41 +08:00
peng xiao
9cfa24e515 Fix thread sync for csbp. 2013-04-23 17:35:40 +08:00
yao
b386ea72aa use float when sum overflow 2013-04-23 17:23:05 +08:00
peng xiao
3282aaa1f5 Fix a bug found on Intel OpenCL SDK. 2013-04-23 15:34:40 +08:00
Peng Xiao
c701d54281 Fix build errors 2013-04-21 11:19:37 +08:00
Peng Xiao
7b08d5ec69 Add OpenCL stereo CSBP implementation 2013-04-20 00:34:37 +08:00
Andrey Kamaev
28aefc4f5a Merge pull request #817 from pengx17:2.4_ocl_bitwise_cleanup 2013-04-17 15:45:38 +04:00
Andrey Kamaev
f6848b66d6 Merge pull request #826 from pengx17:2.4_canny_clampfix 2013-04-17 11:09:49 +04:00
peng xiao
0f7d7100e5 Add clamping for y dimension. 2013-04-16 15:49:15 +08:00
peng xiao
6f63a5d8d5 Let wave_size=1 on CPU. 2013-04-16 14:41:30 +08:00
yao
69a0b5dde5 Add OclCascadeClassifierBuf interface 2013-04-15 17:24:30 +08:00
yao
dec6a3b080 make boxfilter kernel compile on Mac GPU OCL 2013-04-15 16:46:25 +08:00
peng xiao
06a4bad809 Merge ocl and/or/xor operators into one kernel each. 2013-04-15 10:36:03 +08:00
Peng Xiao
6dd6013546 Use a faster way to count 1's (used by Hamming) 2013-04-13 14:34:30 +08:00
Peng Xiao
6b6b1c9cbf Allow more query/train types for ocl::bfmatcher 2013-04-13 14:08:58 +08:00
Peng Xiao
1db20099a9 Enable runtime type definition in kernels 2013-04-13 12:50:17 +08:00
Peng Xiao
fd1528795e Pass query type T into kernel 2013-04-13 11:39:13 +08:00
Peng Xiao
63813e83ae Untabify 2013-04-13 11:22:22 +08:00
Andrey Kamaev
36028bd8ad Merge pull request #808 from bitwangyaoyao:2.4_mac 2013-04-12 14:59:45 +04:00
Vadim Pisarevsky
03e2a52e2c Merge pull request #807 from pengx17:2.4_ocl_bfm_opt 2013-04-12 13:46:55 +04:00
Vadim Pisarevsky
a770d04e78 Merge pull request #781 from bitwangyaoyao:2.4_fixerr 2013-04-12 13:44:59 +04:00
yao
719e8674ad fix the compile errors on Mac 2013-04-12 17:38:59 +08:00
peng xiao
2338a895f5 Capitalize macro namings. 2013-04-12 16:56:49 +08:00
peng xiao
c9d8eb7a84 Fix build error on linux. 2013-04-12 16:52:21 +08:00
peng xiao
6eefd276cf Further optimize bfmatcher by passing macros. 2013-04-12 16:51:36 +08:00
peng xiao
113b7584e0 Optimize bfmatcher by passing macros. 2013-04-12 16:50:30 +08:00
peng xiao
1e49c00f4b Replace create with ensureSizeIsEnough thus buffer objects can be reused. 2013-04-12 16:47:44 +08:00
Andrey Kamaev
9ba25e9d09 Merge pull request #742 from bitwangyaoyao:2.4_fix 2013-04-11 11:33:45 +04:00
yao
7726e273a9 merge add and sub into one set of kernels 2013-04-06 13:37:36 +08:00
yao
bee970ab94 remove the C3 kernels in arithm, as the oclMat will never store 3 channels data 2013-04-05 21:29:29 +08:00
Andrey Kamaev
72e49dc790 Merge pull request #777 from taka-no-me:ocl_win_hang 2013-04-05 13:00:03 +04:00
Andrey Kamaev
a2d27429e4 Merge pull request #775 from bitwangyaoyao:2.4_fixerr 2013-04-05 12:59:45 +04:00
Andrey Kamaev
3400d83778 Workaround hanging of ocl module when ocl API in not really used 2013-04-05 11:01:28 +04:00
yao
5022bc8c25 move the "cpu device checking" from supportsFeatures() to queryDeviceInfo() 2013-04-05 09:17:14 +08:00
yao
bcc086baa9 fix all redefine build errors on some Intel OCL 2013-04-05 08:15:05 +08:00
peng xiao
917138f565 Fix compilation errors. 2013-04-03 17:36:05 +08:00
peng xiao
7758322fd3 Fix some build errors. 2013-04-03 16:42:44 +08:00
peng xiao
ecea583afd Add ocl::stereobp function.
OpenCL StereoBeliefPropagation, ported from GPU implementation.
2013-04-03 15:57:26 +08:00
yao
d5aaea2749 fix some mismatch on cpu device running OCL 2013-04-03 14:24:55 +08:00
yao
cb63bbf001 fix hog on some CPU device running ocl 2013-04-03 13:58:44 +08:00
yao
fd4a6f0af0 make the sparse method give correct results on CPU ocl
Add CL_CPU to supportsFeature check
simplify the logic of pyrlk
2013-04-03 13:23:04 +08:00
Andrey Kamaev
656594ad4f Merge pull request #736 from bitwangyaoyao:2.4_clflush 2013-04-02 15:25:19 +04:00
yao
f3254b28f2 use clflush replaces clfinish 2013-04-02 14:41:02 +08:00
yao
10f6ebfdf7 fix the crash when calling dft 2013-04-02 11:35:40 +08:00
Andrey Kamaev
56d62118d5 Merge pull request #707 from pengx17:2.4_surf 2013-03-29 18:20:45 +04:00
peng xiao
04a6ab4144 Fix OCL Canny 2013-03-27 15:08:51 +08:00
peng xiao
8ffc15371d Fix compiler errors 2013-03-26 17:23:38 +08:00
peng xiao
9698079ca5 Pass warp size into SURF 2013-03-26 15:48:15 +08:00
peng xiao
0c19a07bf4 Add a function to query ocl device info
Currently the function only supports wavefront size query
2013-03-26 15:36:49 +08:00
yao
d6f1ad8c14 more fix 2013-03-26 14:23:26 +08:00
yao
f36db3a037 more fix of mismatch 2013-03-26 14:10:29 +08:00
yao
ad6aae4583 more fix of mismatch functions on CPU OCL 2013-03-26 13:41:13 +08:00
yao
2c06e59a69 fix some mismatch 2013-03-26 13:05:01 +08:00
Andrey Kamaev
4bd721ad3b Fix build errors 2013-03-21 18:37:34 +04:00
Andrey Kamaev
7b8ad4cb04 Refactor OpenCL initialization and allow to use ocl module witout explicit setup 2013-03-21 18:18:15 +04:00
Andrey Kamaev
dd678121b3 Trying to make ocl surf work
1. Added more sync to reduction.
2. Turned off Image2D feature. Probably its support is not detected correctly.
3. Temporary disabled descriptor tests - can't localize a problem of the ocl descriptor.
2013-03-21 18:16:59 +04:00
Andrey Kamaev
1be58f9a00 SURF accuracy test is moved to nonfree 2013-03-21 18:15:46 +04:00
Andrey Kamaev
6846f881a2 Move OpenCL SURF to nonfree module 2013-03-21 17:59:35 +04:00
Andrey Kamaev
91ac9688a8 Allow OpenCL acceleration in every OpenCV module 2013-03-21 17:57:01 +04:00
Andrey Kamaev
00d8ad9e7e Merge pull request #632 from pengx17:2.4 2013-03-15 10:34:43 +04:00
Andrey Kamaev
18ca645fd2 Merge pull request #631 from bitwangyaoyao:2.4_stereo 2013-03-15 10:34:27 +04:00
Andrey Kamaev
ecf770d49d Merge pull request #638 from bitwangyaoyao:2.4_fixErode 2013-03-14 11:41:02 +04:00
Andrey Kamaev
5f3eb0fc49 Merge pull request #628 from bitwangyaoyao:2.4_fixMoments 2013-03-14 11:40:47 +04:00
yao
084385cf38 Fix a bug in ocl::Erode/Dilate, simplify the host logic 2013-03-12 16:45:43 +08:00