Alexander Alekhin [Wed, 29 Nov 2017 16:32:31 +0000 (16:32 +0000)]
Merge pull request #10193 from alalek:fix_ios_cmake_3.10
Alexander Alekhin [Wed, 29 Nov 2017 12:57:32 +0000 (15:57 +0300)]
cmake: fix iOS build with CMake 3.10
Alexander Alekhin [Wed, 29 Nov 2017 12:05:10 +0000 (12:05 +0000)]
Merge pull request #10164 from pengli:dnn
Vadim Pisarevsky [Wed, 29 Nov 2017 11:42:22 +0000 (11:42 +0000)]
Merge pull request #10180 from alalek:ocl_avoid_unnecessary_initialization
Vadim Pisarevsky [Wed, 29 Nov 2017 10:01:58 +0000 (10:01 +0000)]
Merge pull request #10182 from romanc:fix/divByZeroSimdDouble
Vadim Pisarevsky [Wed, 29 Nov 2017 09:51:20 +0000 (09:51 +0000)]
Merge pull request #10170 from LaurentBerger:Issue10166
Alexander Alekhin [Tue, 28 Nov 2017 18:42:17 +0000 (18:42 +0000)]
Merge pull request #10185 from alalek:fix_build_xcode_9.1
Alexander Alekhin [Tue, 28 Nov 2017 14:03:40 +0000 (17:03 +0300)]
build: eliminate Xcode 9.1 warnings
- 3rdparty/libjasper/jas_getopt.c:129:25: warning: this function declaration is not a prototype [-Wstrict-prototypes]
- clang: warning: optimization flag '-fomit-frame-pointer' is not supported for target 'armv7' [-Wignored-optimization-argument]
Vadim Pisarevsky [Tue, 28 Nov 2017 13:19:05 +0000 (13:19 +0000)]
Merge pull request #9986 from terfendail:drawcontours_docfix
Vadim Pisarevsky [Tue, 28 Nov 2017 13:17:43 +0000 (13:17 +0000)]
Merge pull request #10141 from LaurentBerger:MLP_ReLU
Vadim Pisarevsky [Tue, 28 Nov 2017 13:17:10 +0000 (13:17 +0000)]
Merge pull request #10149 from mshabunin:fix-saturate-intrin
Roman Cattaneo [Tue, 28 Nov 2017 12:48:40 +0000 (13:48 +0100)]
Fixes Issue #10181
This PR fixes incorrect division by zero handling in template
specialization of `Div_SIMD` for type `double`.
Alexander Alekhin [Fri, 24 Nov 2017 14:34:02 +0000 (17:34 +0300)]
ocl: avoid unnecessary loading/initializing OpenCL subsystem
If there are no OpenCL/UMat methods calls from application.
OpenCL subsystem is initialized:
- haveOpenCL() is called from application
- useOpenCL() is called from application
- access to OpenCL allocator: UMat is created (empty UMat is ignored) or UMat <-> Mat conversions are called
Don't call OpenCL functions if OPENCV_OPENCL_RUNTIME=disabled
(independent from OpenCL linkage type)
Vadim Pisarevsky [Tue, 28 Nov 2017 10:53:00 +0000 (10:53 +0000)]
Merge pull request #10124 from zhongwuzw:master
Vadim Pisarevsky [Tue, 28 Nov 2017 10:51:06 +0000 (10:51 +0000)]
Merge pull request #10173 from dkurt:update_priorbox_layer
Vadim Pisarevsky [Tue, 28 Nov 2017 10:46:41 +0000 (10:46 +0000)]
Merge pull request #10152 from alalek:py_fix_isalgorithm
Alexander Alekhin [Tue, 28 Nov 2017 10:35:33 +0000 (10:35 +0000)]
Merge pull request #9247 from paroj:wrap_alog_rw
Maksim Shabunin [Thu, 23 Nov 2017 14:58:34 +0000 (17:58 +0300)]
Universal Intrinsics: aligned v_pack behavior on different platforms, fixed 64-bit register on ARM, added more saturate_cast variants
LaurentBerger [Wed, 22 Nov 2017 21:07:23 +0000 (22:07 +0100)]
Add ReLU and LeakyReLU activation function in ml module
Alexander Alekhin [Tue, 28 Nov 2017 09:17:42 +0000 (09:17 +0000)]
Merge pull request #10177 from alalek:cmake_update_optimization_filter
Alexander Alekhin [Tue, 28 Nov 2017 05:39:29 +0000 (05:39 +0000)]
Merge pull request #10178 from alalek:ts_message
Alexander Alekhin [Fri, 13 Oct 2017 13:08:51 +0000 (16:08 +0300)]
ts: detailed message
Pavel Rojtberg [Fri, 4 Aug 2017 15:12:20 +0000 (17:12 +0200)]
add test_algorithm_rw using AKAZE
Pavel Rojtberg [Thu, 27 Jul 2017 12:34:32 +0000 (14:34 +0200)]
Python: wrap Algorithm::read and Algorithm::write
Alexander Alekhin [Mon, 27 Nov 2017 15:16:01 +0000 (18:16 +0300)]
cmake: update optimization filter
Wu Zhiwen [Mon, 27 Nov 2017 04:57:36 +0000 (12:57 +0800)]
dnn(ocl4dnn): fuseLayer() use umat_input/outputBlobs for OpenCL target
Also, fix bug when use OPENCL target but no OpenCL runtime
Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com>
Li Peng [Fri, 24 Nov 2017 10:22:59 +0000 (18:22 +0800)]
Add ocl accuracy test for a few dnn nets
They are alexnet, mobilenet-ssd, resnet50, squeezeNet_v1_1,
yolo and fast_neural_style.
Signed-off-by: Li Peng <peng.li@intel.com>
Dmitry Kurtaev [Mon, 27 Nov 2017 12:59:31 +0000 (15:59 +0300)]
Update PriorBox layer
LaurentBerger [Mon, 27 Nov 2017 12:13:05 +0000 (13:13 +0100)]
Try to solve issue 10166
Alexander Alekhin [Mon, 27 Nov 2017 11:05:07 +0000 (11:05 +0000)]
Merge pull request #10167 from alalek:ocl_fix_issue_contrib1467
Vadim Pisarevsky [Mon, 27 Nov 2017 10:42:10 +0000 (13:42 +0300)]
Update imgproc.hpp
rewrote the added note in perhaps a cleaner way
Vadim Pisarevsky [Mon, 27 Nov 2017 10:12:15 +0000 (10:12 +0000)]
Merge pull request #10107 from ElenaGvozdeva:medianBlur_HAL
Vadim Pisarevsky [Mon, 27 Nov 2017 10:04:03 +0000 (10:04 +0000)]
Merge pull request #10154 from alalek:ocl_cleanup_obsolete_cache
Vadim Pisarevsky [Mon, 27 Nov 2017 10:03:31 +0000 (10:03 +0000)]
Merge pull request #10155 from alalek:ocl_update_loader
Alexander Alekhin [Mon, 27 Nov 2017 09:43:29 +0000 (12:43 +0300)]
ocl: fix null pointer access crash
Alexander Alekhin [Mon, 27 Nov 2017 09:24:49 +0000 (09:24 +0000)]
Merge pull request #10156 from dkurt:refresh_torch_enet_sample
Alexander Alekhin [Mon, 27 Nov 2017 07:56:06 +0000 (07:56 +0000)]
Merge pull request #10163 from windhw:patch-1
Wei Hao [Mon, 27 Nov 2017 06:36:51 +0000 (14:36 +0800)]
Update houghlines.py
Fix a code indent bug: cv2.imshow("detected lines", cdst)
Alexander Alekhin [Sun, 26 Nov 2017 12:00:16 +0000 (12:00 +0000)]
Merge pull request #10055 from mabdullahrafique:patch-1
Alexander Alekhin [Sat, 25 Nov 2017 12:31:47 +0000 (12:31 +0000)]
Merge pull request #10095 from alalek:fix_canny_intrinsics
Dmitry Kurtaev [Fri, 24 Nov 2017 18:20:18 +0000 (21:20 +0300)]
Update Torch ENet sample
Alexander Alekhin [Fri, 24 Nov 2017 16:09:05 +0000 (19:09 +0300)]
ocl: update OpenCL runtime loader
- fix compilation on Apple (undefined 'oclpath')
- don't warn on OPENCV_OPENCL_RUNTIME=disabled
Alexander Alekhin [Fri, 24 Nov 2017 16:17:28 +0000 (16:17 +0000)]
Merge pull request #10153 from dkurt:specify_mobilenet_version
Alexander Alekhin [Fri, 24 Nov 2017 09:52:29 +0000 (12:52 +0300)]
ocl: drop obsolete cache directories after upgrade of OpenCL driver
Entries with the same platform name, the same device name and with different driver versions
are assumed obsolete.
Vadim Pisarevsky [Fri, 24 Nov 2017 11:51:03 +0000 (11:51 +0000)]
Merge pull request #10150 from alalek:dnn_ocl_refactor_pooling
Dmitry Kurtaev [Fri, 24 Nov 2017 10:40:35 +0000 (13:40 +0300)]
Specific version of MobileNet-SSD from TensorFlow
Alexander Alekhin [Mon, 28 Aug 2017 12:29:33 +0000 (15:29 +0300)]
bindings(py): fix handling of 'isalgorithm'
Alexander Alekhin [Thu, 23 Nov 2017 20:45:44 +0000 (20:45 +0000)]
dnn(ocl4dnn): drop unused batch_size_ in pooling
Alexander Alekhin [Thu, 23 Nov 2017 20:10:53 +0000 (20:10 +0000)]
dnn(ocl4dnn): refactor pooling OpenCL calls
Muhammad Abdullah [Wed, 8 Nov 2017 23:46:35 +0000 (15:46 -0800)]
update digits_video.py
Following were the errors in the digits_video.py
1 ) The code was not working because data type of x was float however in "cv2.rectangle" we require integer .
2 ) After pressing the "esc" button the image windows did not destroy
So I amended following things:
1 ) ~converted data type of x to int.~ Used Python integer division (//)
2 ) used cv2.destroyAllWindows() to close all windows after the press of "esc" by user.
Alexander Alekhin [Thu, 23 Nov 2017 18:47:14 +0000 (18:47 +0000)]
Merge pull request #10143 from pengli:ocl4dnn
Maksim Shabunin [Thu, 23 Nov 2017 18:34:59 +0000 (18:34 +0000)]
Merge pull request #10148 from alalek:fix_ffmpeg_sws_scale_crash
Arthur Pastel [Thu, 23 Nov 2017 18:27:11 +0000 (19:27 +0100)]
Merge pull request #9895 from art049:parrallel_calibration_py
* Adding threading in calibrate.py
* samples: update calibrate.py
Alexander Alekhin [Thu, 23 Nov 2017 18:22:37 +0000 (18:22 +0000)]
Merge pull request #9996 from dkurt:dnn_multiple_inputs
Li Peng [Thu, 16 Nov 2017 03:20:08 +0000 (11:20 +0800)]
use OutputArrayOfArrays in net forward interface
It allows umat buffers used in net forward interface
Signed-off-by: Li Peng <peng.li@intel.com>
Wu, Zhiwen [Tue, 31 Oct 2017 01:21:08 +0000 (09:21 +0800)]
cleanup ocl4dnn spatial convolution kernels
remove unused macros and half definition macros,
also remove unused ocl::Queue
Signed-off-by: Li Peng <peng.li@intel.com>
Alexander Alekhin [Thu, 23 Nov 2017 14:48:01 +0000 (17:48 +0300)]
videoio(ffmpeg): more workarounds for sws_scale() crash
Prevents access data after the end of input buffer near 4K page boundaries
Alexander Alekhin [Thu, 23 Nov 2017 15:09:35 +0000 (15:09 +0000)]
Merge pull request #9893 from mshabunin:add-modules-whitelist
Alexander Alekhin [Thu, 23 Nov 2017 14:10:42 +0000 (14:10 +0000)]
Merge pull request #10139 from alalek:dnn_rename_caffe_proto_package
Alexander Alekhin [Thu, 23 Nov 2017 13:37:32 +0000 (13:37 +0000)]
Merge pull request #10018 from alalek:ocl_binary_cache
Maksim Shabunin [Thu, 19 Oct 2017 14:37:36 +0000 (17:37 +0300)]
Added BUILD_LIST feature
Vadim Pisarevsky [Thu, 23 Nov 2017 12:51:55 +0000 (12:51 +0000)]
Merge pull request #10142 from alalek:pytests_update
Vadim Pisarevsky [Thu, 23 Nov 2017 12:50:25 +0000 (12:50 +0000)]
Merge pull request #10144 from mshabunin:fix-videocapture-open-order
Maksim Shabunin [Thu, 23 Nov 2017 10:31:07 +0000 (13:31 +0300)]
videoio: fixed capture open order, added apiPrefernce to FFMPEG tests
elenagvo [Thu, 23 Nov 2017 10:30:00 +0000 (13:30 +0300)]
fix the parameters order
Alexander Alekhin [Thu, 23 Nov 2017 10:08:26 +0000 (13:08 +0300)]
python(test): don't write bytecode
Tests are usually lauched from source directory, so additional unnecessary
files should be eliminated.
Alternative ways (command line):
- python -B ...
- PYTHONDONTWRITEBYTECODE=1 python ...
Alexander Alekhin [Mon, 28 Aug 2017 14:26:49 +0000 (17:26 +0300)]
python(test): add tests filter
Usage example (bash):
$ OPENCV_PYTEST_FILTER=test_digits python test.py -v
Alexander Alekhin [Wed, 22 Nov 2017 18:39:47 +0000 (18:39 +0000)]
Merge pull request #10136 from alalek:issue_10134
Alexander Alekhin [Wed, 22 Nov 2017 18:15:00 +0000 (18:15 +0000)]
Merge pull request #9994 from r2d3:dnn_memory_load
Alexander Alekhin [Wed, 22 Nov 2017 18:03:51 +0000 (18:03 +0000)]
Merge pull request #10126 from alalek:dnn_issue_10125
Alexander Alekhin [Wed, 22 Nov 2017 15:31:55 +0000 (18:31 +0300)]
dnn: autogenerated files
Alexander Alekhin [Wed, 22 Nov 2017 15:31:36 +0000 (18:31 +0300)]
dnn: rename caffe protobuf package
Alexander Alekhin [Wed, 22 Nov 2017 12:52:40 +0000 (15:52 +0300)]
core: fix compilation of intrinsic code
Alexander Alekhin [Wed, 22 Nov 2017 11:00:58 +0000 (14:00 +0300)]
dnn: sync output/internals blobs back
Alexander Alekhin [Thu, 12 Oct 2017 11:23:45 +0000 (14:23 +0300)]
ocl: binary program cache
Vadim Pisarevsky [Wed, 22 Nov 2017 08:16:20 +0000 (08:16 +0000)]
Merge pull request #10092 from alalek:dnn_rename_caffe_proto
Vadim Pisarevsky [Wed, 22 Nov 2017 08:13:35 +0000 (08:13 +0000)]
Merge pull request #9931 from mshabunin:bss-cleanup
Maksim Shabunin [Tue, 21 Nov 2017 15:07:30 +0000 (18:07 +0300)]
Fixed allocSingleton
Maksim Shabunin [Tue, 21 Nov 2017 14:54:57 +0000 (17:54 +0300)]
align singleton malloc
Maksim Shabunin [Tue, 21 Nov 2017 14:54:43 +0000 (17:54 +0300)]
Other buffers
Maksim Shabunin [Tue, 21 Nov 2017 14:34:10 +0000 (17:34 +0300)]
init ABtoXZ
Maksim Shabunin [Tue, 21 Nov 2017 14:28:45 +0000 (17:28 +0300)]
LUV tables
Maksim Shabunin [Tue, 21 Nov 2017 13:49:58 +0000 (16:49 +0300)]
static init
Maksim Shabunin [Mon, 23 Oct 2017 10:48:14 +0000 (13:48 +0300)]
cvtColor: fixed tables init, moved some tables to heap
Alexander Alekhin [Tue, 21 Nov 2017 14:28:42 +0000 (17:28 +0300)]
dnn: drop OpenCL code path for DetectionOutputLayer
getUMat()/getMat() calls are scope based. Results of these calls can't be
stored somewhere for future usage.
Vadim Pisarevsky [Tue, 21 Nov 2017 12:40:19 +0000 (12:40 +0000)]
Merge pull request #10121 from alalek:ocl_syntax_fix
Rostislav Vasilikhin [Tue, 21 Nov 2017 11:18:47 +0000 (14:18 +0300)]
Merge pull request #10041 from savuor:RevHoughWorks
HoughCircles rewritten (PR #7434 updated) (#10041)
* initial version of renewed HoughCircles done
* fixed compilation
* fixed SIMD ability & compilation warning
* fixed accumulator nonmax comparison
* common Mutex for all invokers
* nzLocal is std::vector
* nz is std::vector
* SSE2 -> SIMD128
* centers is now std::vector
* circles is std::vector
* estimateRadius updated
* accum calculation w/o mutex
* less deprecated code
* several bugs fixed
* back to mutex, TLS gathering doesn't work
* extra code removed
* little refactoring
* docs note updated
* a little speedup
* warning fixed
zhongwuzw [Tue, 21 Nov 2017 04:56:23 +0000 (12:56 +0800)]
Fix customPreviewLayer layout issue when rotateVideo is NO in iOS
Alexander Alekhin [Mon, 20 Nov 2017 18:13:33 +0000 (18:13 +0000)]
Merge pull request #10113 from wzw-intel:fusion
Alexander Alekhin [Mon, 20 Nov 2017 16:19:35 +0000 (19:19 +0300)]
core(ocl): syntax fix
Alexander Alekhin [Wed, 15 Nov 2017 11:07:41 +0000 (14:07 +0300)]
dnn: autogenerated files
rename caffe.proto => opencv-caffe.proto
Alexander Alekhin [Wed, 15 Nov 2017 11:07:41 +0000 (14:07 +0300)]
dnn: avoid conflicts with original caffe.proto
rename caffe.proto => opencv-caffe.proto
Alexander Alekhin [Mon, 20 Nov 2017 15:59:56 +0000 (15:59 +0000)]
Merge pull request #10120 from dkurt:remove_caffe_header_from_layer
David Geldreich [Fri, 27 Oct 2017 16:01:41 +0000 (18:01 +0200)]
add loading TensorFlow/Caffe net from memory buffer
add a corresponding test
Dmitry Kurtaev [Mon, 20 Nov 2017 15:01:40 +0000 (18:01 +0300)]
Replace caffe::NormalizedBBox to local structure
elenagvo [Fri, 17 Nov 2017 13:21:35 +0000 (16:21 +0300)]
add HAL for medianBlur
Tomoaki Teshima [Mon, 20 Nov 2017 12:56:53 +0000 (21:56 +0900)]
Merge pull request #9753 from tomoaki0705:universalMatmul
* add accuracy test and performance check for matmul
* add performance tests for transform and dotProduct
* add test Core_TransformLargeTest for 8u version of transform
* remove raw SSE2/NEON implementation from matmul.cpp
* use universal intrinsic instead of raw intrinsic
* remove unused templated function
* add v_matmuladd which multiply 3x3 matrix and add 3x1 vector
* add v_rotate_left/right in universal intrinsic
* suppress intrinsic on some function and platform
* add pure SW implementation of new universal intrinsics
* add test for new universal intrinsics
* core: prevent memory access after the end of buffer
* fix perf tests
gdkessler [Mon, 20 Nov 2017 12:07:24 +0000 (07:07 -0500)]
Merge pull request #10093 from gdkessler/gdal_image_read_fix_10089
Fix GDAL image decoding color problems identified by issue #10089, by: (#10093)
* Fix GDAL image decoding color problems identified by issue #10089, by:
Fixing CV_8UC1 symbol, which should be CV_8UC3 for RGB GDAL color table images.
Fixing image.ptr<VecX>(row,col)[] to be (*image.ptr<VecX>(row,col))[] to correctly access VecX array elements, as ptr<VecX>() returns a pointer to the VecX, not the first element of VecX. This fixes the color problem with color table gif images, and avoids out-of-bounds memory access.
Respecting the color identification of raster bands provided by the GDAL image driver, and produce BGR or BGRA images. Note that color bands of images using the HSL, CMY, CMYK, or YCbCr color space are ignored, rather than converting them to BGR.
* When reading image files using the GDAL decoder, exit with an error if a color band is encountered that isn't used (eg. from CMYK or YCbCbr), rather than silently ignoring the band's data.
Wu Zhiwen [Mon, 20 Nov 2017 03:29:18 +0000 (11:29 +0800)]
dnn(ocl4dnn): add fusion support for Power activation and eltwise add
Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com>
Alexander Alekhin [Sat, 18 Nov 2017 18:07:14 +0000 (18:07 +0000)]
Merge pull request #10087 from mshabunin:small-cmake-fixes