Vadim Pisarevsky [Wed, 11 Oct 2017 17:11:15 +0000 (17:11 +0000)]
Merge pull request #9824 from sturkmen72:upd_minEnclosingTriangle
Alexander Alekhin [Wed, 11 Oct 2017 15:21:05 +0000 (15:21 +0000)]
Merge pull request #9836 from lupustr3:pvlasov/external_ipp-iw_fix
Alexander Alekhin [Wed, 11 Oct 2017 15:19:17 +0000 (15:19 +0000)]
Merge pull request #9818 from tz70s:issue#9570
Suleyman TURKMEN [Wed, 11 Oct 2017 14:52:23 +0000 (17:52 +0300)]
Update minarea.cpp
Suleyman TURKMEN [Wed, 11 Oct 2017 14:48:19 +0000 (17:48 +0300)]
Update min_enclosing_triangle.cpp
Suleyman TURKMEN [Wed, 11 Oct 2017 14:37:38 +0000 (17:37 +0300)]
Update imgproc.hpp
Vadim Pisarevsky [Wed, 11 Oct 2017 13:36:14 +0000 (13:36 +0000)]
Merge pull request #9636 from dkurt:duplicate_lp_norm_layer
Vadim Pisarevsky [Wed, 11 Oct 2017 13:31:46 +0000 (13:31 +0000)]
Merge pull request #9820 from sovrasov:text_detector_dnn
Vadim Pisarevsky [Wed, 11 Oct 2017 13:31:17 +0000 (13:31 +0000)]
Merge pull request #9828 from berak:fix_c++17_9572
Vadim Pisarevsky [Wed, 11 Oct 2017 13:28:38 +0000 (13:28 +0000)]
Merge pull request #9823 from alalek:dnn_halide_bypass_tbb_threads
Pavel Vlasov [Wed, 11 Oct 2017 12:30:01 +0000 (15:30 +0300)]
Standalone IPP with IW fix;
Vladislav Sovrasov [Wed, 11 Oct 2017 11:04:48 +0000 (14:04 +0300)]
dnn: fix additional text boxes handling after the latest adaptations for TF
Alexander Alekhin [Wed, 11 Oct 2017 10:58:04 +0000 (10:58 +0000)]
Merge pull request #9827 from ryanfox:patch-2
Alexander Alekhin [Wed, 11 Oct 2017 08:55:15 +0000 (08:55 +0000)]
Merge pull request #9826 from p0wdrdotcom:master
Vladislav Sovrasov [Tue, 10 Oct 2017 09:03:05 +0000 (12:03 +0300)]
dnn: modify priorBox layer
berak [Wed, 11 Oct 2017 07:50:22 +0000 (09:50 +0200)]
fix c++17 namsespace issues
Ryan Fox [Wed, 11 Oct 2017 02:37:26 +0000 (21:37 -0500)]
fix some grammatical errors
Geoff McIver [Tue, 10 Oct 2017 20:32:38 +0000 (09:32 +1300)]
This statement was keeping HAAR cascades from leveraging opencl on nvidia devices. "localSize" on the featureEvaluator remains Size(0, 0) which sets the bool "use_ocl" to false. Adding this allows NVidia GPUs to leverage opencl HAAR Cascades
Suleyman TURKMEN [Sun, 8 Oct 2017 10:22:13 +0000 (13:22 +0300)]
Updates min_enclosing_triangle.cpp
Dmitry Kurtaev [Mon, 28 Aug 2017 14:37:09 +0000 (17:37 +0300)]
Removed LPNormalize layer.
Alexander Alekhin [Tue, 10 Oct 2017 14:52:55 +0000 (17:52 +0300)]
dnn(halide): don't compile Halide via parallel_for_()
To avoid problem with reduced stack size of inner threads.
Vadim Pisarevsky [Tue, 10 Oct 2017 14:23:02 +0000 (14:23 +0000)]
Merge pull request #9816 from opalmirror:fix_stereobm_mindisp_truncation_1
Vadim Pisarevsky [Tue, 10 Oct 2017 14:22:00 +0000 (14:22 +0000)]
Merge pull request #9822 from alalek:fix_tbb_search
Vadim Pisarevsky [Tue, 10 Oct 2017 13:51:39 +0000 (13:51 +0000)]
Merge pull request #8940 from 678098:nonblocking_haar_detector_parallel_for
tz70s [Tue, 10 Oct 2017 05:28:07 +0000 (13:28 +0800)]
fix#9570: implement mat ptr for generic types
The original template based mat ptr for indexing is not implemented,
add the similar implementation as uchar type, but cast to
user-defined type from the uchar pointer.
Vadim Pisarevsky [Tue, 10 Oct 2017 12:02:03 +0000 (12:02 +0000)]
Merge pull request #9705 from AlexeyAB:dnn_darknet_yolo_v2
Alexander Alekhin [Tue, 10 Oct 2017 11:18:28 +0000 (14:18 +0300)]
tbb: don't search library in ENV{LD_LIBRARY_PATH}
ENV{LIBRARY_PATH} is enough
Vadim Pisarevsky [Tue, 10 Oct 2017 10:58:34 +0000 (10:58 +0000)]
Merge pull request #9786 from LaurentBerger:Histo3d
Vadim Pisarevsky [Tue, 10 Oct 2017 10:06:24 +0000 (10:06 +0000)]
Merge pull request #9750 from dkurt:feature_dnn_tf_text_graph
Vadim Pisarevsky [Tue, 10 Oct 2017 10:00:31 +0000 (10:00 +0000)]
Merge pull request #9763 from seiko2plus:addVsxCore
Vadim Pisarevsky [Tue, 10 Oct 2017 09:57:51 +0000 (09:57 +0000)]
Merge pull request #9811 from dkurt:prelu_with_shared_channels
Maksim Shabunin [Tue, 10 Oct 2017 09:02:02 +0000 (09:02 +0000)]
Merge pull request #9805 from alalek:ffmpeg_update
James Perkins [Thu, 5 Oct 2017 22:42:20 +0000 (15:42 -0700)]
fix StereoBM disparity map right margin truncation when minDisparities > 0
LaurentBerger [Mon, 9 Oct 2017 20:25:57 +0000 (22:25 +0200)]
It's done
Alexander Alekhin [Mon, 9 Oct 2017 20:08:00 +0000 (20:08 +0000)]
Merge pull request #9815 from alalek:issue_9813
Alexander Alekhin [Mon, 9 Oct 2017 19:12:52 +0000 (22:12 +0300)]
qt: fix non-world builds
AlexeyAB [Sun, 24 Sep 2017 20:34:08 +0000 (23:34 +0300)]
Added DNN Darknet Yolo v2 for object detection
Dmitry Kurtaev [Mon, 9 Oct 2017 14:47:46 +0000 (17:47 +0300)]
PReLU layer from Caffe
Alexander Alekhin [Mon, 9 Oct 2017 16:19:09 +0000 (16:19 +0000)]
Merge pull request #9810 from alalek:fix_js_build
Alexander Alekhin [Mon, 9 Oct 2017 13:14:01 +0000 (16:14 +0300)]
js: skip 'utils' headers
Vadim Pisarevsky [Mon, 9 Oct 2017 12:46:06 +0000 (12:46 +0000)]
Merge pull request #9792 from alalek:port_9776
Vadim Pisarevsky [Mon, 9 Oct 2017 12:43:46 +0000 (12:43 +0000)]
Merge pull request #9799 from alalek:ocl_program
Vadim Pisarevsky [Mon, 9 Oct 2017 12:33:08 +0000 (12:33 +0000)]
Merge pull request #9800 from alalek:fix_build_msvs2010
Vadim Pisarevsky [Mon, 9 Oct 2017 12:11:22 +0000 (12:11 +0000)]
Merge pull request #9803 from wzw-intel:ocl_timer
Vadim Pisarevsky [Mon, 9 Oct 2017 12:09:55 +0000 (12:09 +0000)]
Merge pull request #9806 from mshabunin:fix-static-7
Maksim Shabunin [Mon, 9 Oct 2017 10:52:07 +0000 (13:52 +0300)]
Fixed and suppressed warnings produced by GCC 6.3.0
Alexander Alekhin [Mon, 9 Oct 2017 10:46:50 +0000 (10:46 +0000)]
Merge pull request #9790 from mevatron:patch-1
Alexander Alekhin [Mon, 9 Oct 2017 10:17:17 +0000 (13:17 +0300)]
ffmpeg: update 3.3.4
Maksim Shabunin [Mon, 9 Oct 2017 10:33:55 +0000 (13:33 +0300)]
Fixed warnings produced by static analyzer
LaurentBerger [Mon, 9 Oct 2017 08:59:30 +0000 (10:59 +0200)]
Use @snippet
Wu Zhiwen [Mon, 2 Oct 2017 11:22:28 +0000 (19:22 +0800)]
ocl: simplify ocl::Timer
Use clFinish to gurantee commands completed, instead of waiting for events.
Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com>
Sayed Adel [Mon, 2 Oct 2017 22:55:59 +0000 (22:55 +0000)]
Added universal intrinsic for VSX
Sayed Adel [Mon, 2 Oct 2017 22:54:31 +0000 (22:54 +0000)]
Added support for VSX
Alexander Alekhin [Sun, 8 Oct 2017 20:32:22 +0000 (23:32 +0300)]
build: fix MSVS2010
Dmitry Kurtaev [Thu, 28 Sep 2017 13:51:47 +0000 (16:51 +0300)]
Text TensorFlow graphs parsing. MobileNet-SSD for 90 classes.
Alexander Alekhin [Sun, 8 Oct 2017 17:32:49 +0000 (17:32 +0000)]
Merge pull request #9796 from lzx1413:patch-1
Alexander Alekhin [Thu, 7 Sep 2017 13:41:19 +0000 (16:41 +0300)]
ocl: refactor program compilation
Alexander Alekhin [Thu, 7 Sep 2017 13:40:03 +0000 (16:40 +0300)]
ocl: define ProgramSource before program
no changes in code
Alexander Alekhin [Sun, 8 Oct 2017 16:17:40 +0000 (19:17 +0300)]
dnn(samples): fix python syntax (false -> False)
lzx1413 [Sun, 8 Oct 2017 13:10:25 +0000 (21:10 +0800)]
in python, false should be False
Alexander Alekhin [Sun, 8 Oct 2017 09:48:46 +0000 (09:48 +0000)]
Merge pull request #9795 from IgWod:refactor-scalar-to-raw-data
Igor Wodiany [Sat, 7 Oct 2017 18:46:45 +0000 (19:46 +0100)]
Extract code from scalarToRawData
The same code was repeated several time for different data types, so
it was extracted as a templated function to improve maintability and
make a code more clear.
Igor Wodiany [Wed, 4 Oct 2017 21:54:40 +0000 (22:54 +0100)]
Fix a memory leak in the Mat copying constructor
Exception may be rasied inside the body of a copying constructor after
refcount has been increased, and beacause in the case of the exception
destrcutor is never called what causes memory leak. This commit adds a
workaround that calls the release() function before the exception is
thrown outside the contructor.
mevatron [Fri, 6 Oct 2017 21:03:39 +0000 (16:03 -0500)]
Small fix for using Qt 5.x with opencv_world
LaurentBerger [Thu, 5 Oct 2017 08:06:40 +0000 (10:06 +0200)]
3D histogram
Vadim Pisarevsky [Fri, 6 Oct 2017 14:35:42 +0000 (14:35 +0000)]
Merge pull request #9761 from Jazmann:ellipseFitAMS&Direct
Vadim Pisarevsky [Fri, 6 Oct 2017 14:02:58 +0000 (14:02 +0000)]
Merge pull request #9424 from Cartucho:update_imgproc_tutorials
Vadim Pisarevsky [Fri, 6 Oct 2017 13:47:54 +0000 (13:47 +0000)]
Merge pull request #9772 from dkurt:fix_caffe_eltwise_and_fc_layers
Vadim Pisarevsky [Fri, 6 Oct 2017 13:46:50 +0000 (13:46 +0000)]
Merge pull request #9787 from dkurt:feature_dnn_resize_nearest_neighbor
Vadim Pisarevsky [Fri, 6 Oct 2017 11:48:05 +0000 (11:48 +0000)]
Merge pull request #9778 from dkurt:dnn_colorization
Vadim Pisarevsky [Fri, 6 Oct 2017 11:45:35 +0000 (11:45 +0000)]
Merge pull request #9779 from Lightricks:feature/assetslibrary-remove-link
Dmitry Kurtaev [Fri, 6 Oct 2017 11:24:01 +0000 (14:24 +0300)]
Nearest neighbor resize layer
Maksim Shabunin [Fri, 6 Oct 2017 10:44:58 +0000 (10:44 +0000)]
Merge pull request #9773 from shoeffner:feature/doxygen-tag-file
Dmitry Kurtaev [Thu, 5 Oct 2017 10:04:22 +0000 (13:04 +0300)]
Grayscale colorization model (https://github.com/richzhang/colorization) test.
Yaron Inger [Thu, 5 Oct 2017 17:22:56 +0000 (20:22 +0300)]
videoio: remove AssetsLibrary dependency.
The entire AssetsLibrary framework is deprecated since iOS 8.0. The code
used in the camera example code can use UIKit to save videos to the
camera instead, which allows to avoid linking with PhotoKit instead to
prevent increasing the iOS deployment target.
tribta [Sat, 26 Aug 2017 23:02:29 +0000 (00:02 +0100)]
Tutorial Hough Lines
tribta [Sat, 26 Aug 2017 14:11:46 +0000 (15:11 +0100)]
Tutorial Hough Circles
tribta [Fri, 25 Aug 2017 18:44:55 +0000 (19:44 +0100)]
Tutorial Make Border
tribta [Thu, 24 Aug 2017 15:22:27 +0000 (16:22 +0100)]
Tutorial Filter2D
tribta [Wed, 23 Aug 2017 14:37:11 +0000 (15:37 +0100)]
Tutorial Image Pyramids
tribta [Tue, 22 Aug 2017 11:28:23 +0000 (12:28 +0100)]
Tutorial Sobel Derivatives
tribta [Mon, 21 Aug 2017 23:17:09 +0000 (00:17 +0100)]
Tutorial Laplace Operator
tribta [Mon, 21 Aug 2017 17:51:50 +0000 (18:51 +0100)]
Tutorial Morph Lines Detection
tribta [Sun, 20 Aug 2017 21:54:38 +0000 (22:54 +0100)]
Tutorial Hit-or-Miss
tribta [Sun, 20 Aug 2017 16:14:12 +0000 (17:14 +0100)]
Tutorial Smoothing Images
Vadim Pisarevsky [Thu, 5 Oct 2017 10:51:49 +0000 (10:51 +0000)]
Merge pull request #9762 from dkurt:fix_tensorflow_split_layer
Vadim Pisarevsky [Thu, 5 Oct 2017 10:48:20 +0000 (10:48 +0000)]
Merge pull request #9771 from Bisaloo:patch-1
Sebastian Höffner [Wed, 4 Oct 2017 13:52:12 +0000 (15:52 +0200)]
Configuring Doxygen to generate a tag file.
Additionally adds a tutorial on how to use the tag file in other
projects to link back to the OpenCV documentation.
The tutorial is also the only place which links to the tag file as of
now.
For more information on tag files and external documentation, check
https://www.stack.nl/~dimitri/doxygen/manual/external.html
Alexander Alekhin [Wed, 4 Oct 2017 12:49:44 +0000 (12:49 +0000)]
Merge pull request #9769 from pstieber:FixPrivateIncludeInSaturateCast
Dmitry Kurtaev [Wed, 4 Oct 2017 09:23:35 +0000 (12:23 +0300)]
Multidimensional eltwise layer.
Fixed fully-connected layer axis.
Bisaloo [Wed, 4 Oct 2017 09:16:06 +0000 (11:16 +0200)]
Fix 404 url in canny edge detection tutorial
Peter J. Stieber [Wed, 4 Oct 2017 00:47:52 +0000 (17:47 -0700)]
Replace private.cuda.hpp with conditional include of cuda_fp16.h.
Dmitry Kurtaev [Mon, 2 Oct 2017 19:44:42 +0000 (22:44 +0300)]
Fix TensorFlow split layer
Jasper Shemilt [Mon, 2 Oct 2017 16:56:08 +0000 (17:56 +0100)]
Fix Transposed eigenvals and vecs. Didn't notice at first
Jasper Shemilt [Mon, 2 Oct 2017 15:38:41 +0000 (16:38 +0100)]
Adds fitEllipseAMS to imgproc: The Approximate Mean Square (AMS) proposed by Taubin 1991.
Adds fitEllipseDirect to imgproc: The Direct least square (Direct) method by Fitzgibbon1999.
New Tests are included for the methods.
fitEllipseAMS Tests
fitEllipseDirect Tests
Comparative examples are added to fitEllipse.cpp in Samples.
Alexander Alekhin [Mon, 2 Oct 2017 15:26:37 +0000 (15:26 +0000)]
Merge pull request #9754 from sturkmen72:patch-1
Vadim Pisarevsky [Mon, 2 Oct 2017 14:50:46 +0000 (14:50 +0000)]
Merge pull request #9744 from sturkmen72:Update_HOGDescriptor_Documentation
Alexander Alekhin [Mon, 2 Oct 2017 14:47:21 +0000 (14:47 +0000)]
Merge pull request #9596 from paroj:circle_board_params
pengli [Mon, 2 Oct 2017 12:38:00 +0000 (20:38 +0800)]
Merge pull request #9114 from pengli:dnn_rebase
add libdnn acceleration to dnn module (#9114)
* import libdnn code
Signed-off-by: Li Peng <peng.li@intel.com>
* add convolution layer ocl acceleration
Signed-off-by: Li Peng <peng.li@intel.com>
* add pooling layer ocl acceleration
Signed-off-by: Li Peng <peng.li@intel.com>
* add softmax layer ocl acceleration
Signed-off-by: Li Peng <peng.li@intel.com>
* add lrn layer ocl acceleration
Signed-off-by: Li Peng <peng.li@intel.com>
* add innerproduct layer ocl acceleration
Signed-off-by: Li Peng <peng.li@intel.com>
* add HAVE_OPENCL macro
Signed-off-by: Li Peng <peng.li@intel.com>
* fix for convolution ocl
Signed-off-by: Li Peng <peng.li@intel.com>
* enable getUMat() for multi-dimension Mat
Signed-off-by: Li Peng <peng.li@intel.com>
* use getUMat for ocl acceleration
Signed-off-by: Li Peng <peng.li@intel.com>
* use CV_OCL_RUN macro
Signed-off-by: Li Peng <peng.li@intel.com>
* set OPENCL target when it is available
and disable fuseLayer for OCL target for the time being
Signed-off-by: Li Peng <peng.li@intel.com>
* fix innerproduct accuracy test
Signed-off-by: Li Peng <peng.li@intel.com>
* remove trailing space
Signed-off-by: Li Peng <peng.li@intel.com>
* Fixed tensorflow demo bug.
Root cause is that tensorflow has different algorithm with libdnn
to calculate convolution output dimension.
libdnn don't calculate output dimension anymore and just use one
passed in by config.
* split gemm ocl file
split it into gemm_buffer.cl and gemm_image.cl
Signed-off-by: Li Peng <peng.li@intel.com>
* Fix compile failure
Signed-off-by: Li Peng <peng.li@intel.com>
* check env flag for auto tuning
Signed-off-by: Li Peng <peng.li@intel.com>
* switch to new ocl kernels for softmax layer
Signed-off-by: Li Peng <peng.li@intel.com>
* update softmax layer
on some platform subgroup extension may not work well,
fallback to non subgroup ocl acceleration.
Signed-off-by: Li Peng <peng.li@intel.com>
* fallback to cpu path for fc layer with multi output
Signed-off-by: Li Peng <peng.li@intel.com>
* update output message
Signed-off-by: Li Peng <peng.li@intel.com>
* update fully connected layer
fallback to gemm API if libdnn return false
Signed-off-by: Li Peng <peng.li@intel.com>
* Add ReLU OCL implementation
* disable layer fusion for now
Signed-off-by: Li Peng <peng.li@intel.com>
* Add OCL implementation for concat layer
Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com>
* libdnn: update license and copyrights
Also refine libdnn coding style
Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com>
Signed-off-by: Li Peng <peng.li@intel.com>
* DNN: Don't link OpenCL library explicitly
* DNN: Make default preferableTarget to DNN_TARGET_CPU
User should set it to DNN_TARGET_OPENCL explicitly if want to
use OpenCL acceleration.
Also don't fusion when using DNN_TARGET_OPENCL
* DNN: refine coding style
* Add getOpenCLErrorString
* DNN: Use int32_t/uint32_t instread of alias
* Use namespace ocl4dnn to include libdnn things
* remove extra copyTo in softmax ocl path
Signed-off-by: Li Peng <peng.li@intel.com>
* update ReLU layer ocl path
Signed-off-by: Li Peng <peng.li@intel.com>
* Add prefer target property for layer class
It is used to indicate the target for layer forwarding,
either the default CPU target or OCL target.
Signed-off-by: Li Peng <peng.li@intel.com>
* Add cl_event based timer for cv::ocl
* Rename libdnn to ocl4dnn
Signed-off-by: Li Peng <peng.li@intel.com>
Signed-off-by: wzw <zhiwen.wu@intel.com>
* use UMat for ocl4dnn internal buffer
Remove allocateMemory which use clCreateBuffer directly
Signed-off-by: Li Peng <peng.li@intel.com>
Signed-off-by: wzw <zhiwen.wu@intel.com>
* enable buffer gemm in ocl4dnn innerproduct
Signed-off-by: Li Peng <peng.li@intel.com>
* replace int_tp globally for ocl4dnn kernels.
Signed-off-by: wzw <zhiwen.wu@intel.com>
Signed-off-by: Li Peng <peng.li@intel.com>
* create UMat for layer params
Signed-off-by: Li Peng <peng.li@intel.com>
* update sign ocl kernel
Signed-off-by: Li Peng <peng.li@intel.com>
* update image based gemm of inner product layer
Signed-off-by: Li Peng <peng.li@intel.com>
* remove buffer gemm of inner product layer
call cv::gemm API instead
Signed-off-by: Li Peng <peng.li@intel.com>
* change ocl4dnn forward parameter to UMat
Signed-off-by: Li Peng <peng.li@intel.com>
* Refine auto-tuning mechanism.
- Use OPENCV_OCL4DNN_KERNEL_CONFIG_PATH to set cache directory
for fine-tuned kernel configuration.
e.g. export OPENCV_OCL4DNN_KERNEL_CONFIG_PATH=/home/tmp,
the cache directory will be /home/tmp/spatialkernels/ on Linux.
- Define environment OPENCV_OCL4DNN_ENABLE_AUTO_TUNING to enable
auto-tuning.
- OPENCV_OPENCL_ENABLE_PROFILING is only used to enable profiling
for OpenCL command queue. This fix basic kernel get wrong running
time, i.e. 0ms.
- If creating cache directory failed, disable auto-tuning.
* Detect and create cache dir on windows
Signed-off-by: Li Peng <peng.li@intel.com>
* Refine gemm like convolution kernel.
Signed-off-by: Li Peng <peng.li@intel.com>
* Fix redundant swizzleWeights calling when use cached kernel config.
* Fix "out of resource" bug when auto-tuning too many kernels.
* replace cl_mem with UMat in ocl4dnnConvSpatial class
* OCL4DNN: reduce the tuning kernel candidate.
This patch could reduce 75% of the tuning candidates with less
than 2% performance impact for the final result.
Signed-off-by: Zhigang Gong <zhigang.gong@intel.com>
* replace cl_mem with umat in ocl4dnn convolution
Signed-off-by: Li Peng <peng.li@intel.com>
* remove weight_image_ of ocl4dnn inner product
Actually it is unused in the computation
Signed-off-by: Li Peng <peng.li@intel.com>
* Various fixes for ocl4dnn
1. OCL_PERFORMANCE_CHECK(ocl::Device::getDefault().isIntel())
2. Ptr<OCL4DNNInnerProduct<float> > innerProductOp
3. Code comments cleanup
4. ignore check on OCL cpu device
Signed-off-by: Li Peng <peng.li@intel.com>
* add build option for log softmax
Signed-off-by: Li Peng <peng.li@intel.com>
* remove unused ocl kernels in ocl4dnn
Signed-off-by: Li Peng <peng.li@intel.com>
* replace ocl4dnnSet with opencv setTo
Signed-off-by: Li Peng <peng.li@intel.com>
* replace ALIGN with cv::alignSize
Signed-off-by: Li Peng <peng.li@intel.com>
* check kernel build options
Signed-off-by: Li Peng <peng.li@intel.com>
* Handle program compilation fail properly.
* Use std::numeric_limits<float>::infinity() for large float number
* check ocl4dnn kernel compilation result
Signed-off-by: Li Peng <peng.li@intel.com>
* remove unused ctx_id
Signed-off-by: Li Peng <peng.li@intel.com>
* change clEnqueueNDRangeKernel to kernel.run()
Signed-off-by: Li Peng <peng.li@intel.com>
* change cl_mem to UMat in image based gemm
Signed-off-by: Li Peng <peng.li@intel.com>
* check intel subgroup support for lrn and pooling layer
Signed-off-by: Li Peng <peng.li@intel.com>
* Fix convolution bug if group is greater than 1
Signed-off-by: Li Peng <peng.li@intel.com>
* Set default layer preferableTarget to be DNN_TARGET_CPU
Signed-off-by: Li Peng <peng.li@intel.com>
* Add ocl perf test for convolution
Signed-off-by: Li Peng <peng.li@intel.com>
* Add more ocl accuracy test
Signed-off-by: Li Peng <peng.li@intel.com>
* replace cl_image with ocl::Image2D
Signed-off-by: Li Peng <peng.li@intel.com>
* Fix build failure in elementwise layer
Signed-off-by: Li Peng <peng.li@intel.com>
* use getUMat() to get blob data
Signed-off-by: Li Peng <peng.li@intel.com>
* replace cl_mem handle with ocl::KernelArg
Signed-off-by: Li Peng <peng.li@intel.com>
* dnn(build): don't use C++11, OPENCL_LIBRARIES fix
* dnn(ocl4dnn): remove unused OpenCL kernels
* dnn(ocl4dnn): extract OpenCL code into .cl files
* dnn(ocl4dnn): refine auto-tuning
Defaultly disable auto-tuning, set OPENCV_OCL4DNN_ENABLE_AUTO_TUNING
environment variable to enable it.
Use a set of pre-tuned configs as default config if auto-tuning is disabled.
These configs are tuned for Intel GPU with 48/72 EUs, and for googlenet,
AlexNet, ResNet-50
If default config is not suitable, use the first available kernel config
from the candidates. Candidate priority from high to low is gemm like kernel,
IDLF kernel, basick kernel.
* dnn(ocl4dnn): pooling doesn't use OpenCL subgroups
* dnn(ocl4dnn): fix perf test
OpenCV has default 3sec time limit for each performance test.
Warmup OpenCL backend outside of perf measurement loop.
* use ocl::KernelArg as much as possible
Signed-off-by: Li Peng <peng.li@intel.com>
* dnn(ocl4dnn): fix bias bug for gemm like kernel
* dnn(ocl4dnn): wrap cl_mem into UMat
Signed-off-by: Li Peng <peng.li@intel.com>
* dnn(ocl4dnn): Refine signature of kernel config
- Use more readable string as signture of kernel config
- Don't count device name and vendor in signature string
- Default kernel configurations are tuned for Intel GPU with
24/48/72 EUs, and for googlenet, AlexNet, ResNet-50 net model.
* dnn(ocl4dnn): swap width/height in configuration
* dnn(ocl4dnn): enable configs for Intel OpenCL runtime only
* core: make configuration helper functions accessible from non-core modules
* dnn(ocl4dnn): update kernel auto-tuning behavior
Avoid unwanted creation of directories
* dnn(ocl4dnn): simplify kernel to workaround OpenCL compiler crash
* dnn(ocl4dnn): remove redundant code
* dnn(ocl4dnn): Add more clear message for simd size dismatch.
* dnn(ocl4dnn): add const to const argument
Signed-off-by: Li Peng <peng.li@intel.com>
* dnn(ocl4dnn): force compiler use a specific SIMD size for IDLF kernel
* dnn(ocl4dnn): drop unused tuneLocalSize()
* dnn(ocl4dnn): specify OpenCL queue for Timer and convolve() method
* dnn(ocl4dnn): sanitize file names used for cache
* dnn(perf): enable Network tests with OpenCL
* dnn(ocl4dnn/conv): drop computeGlobalSize()
* dnn(ocl4dnn/conv): drop unused fields
* dnn(ocl4dnn/conv): simplify ctor
* dnn(ocl4dnn/conv): refactor kernelConfig localSize=NULL
* dnn(ocl4dnn/conv): drop unsupported double / untested half types
* dnn(ocl4dnn/conv): drop unused variable
* dnn(ocl4dnn/conv): alignSize/divUp
* dnn(ocl4dnn/conv): use enum values
* dnn(ocl4dnn): drop unused innerproduct variable
Signed-off-by: Li Peng <peng.li@intel.com>
* dnn(ocl4dnn): add an generic function to check cl option support
* dnn(ocl4dnn): run softmax subgroup version kernel first
Signed-off-by: Li Peng <peng.li@intel.com>
Pavel Rojtberg [Fri, 8 Sep 2017 14:35:23 +0000 (16:35 +0200)]
calib3d: export parameters of CALIB_CB_CLUSTERING