platform/upstream/opencv.git
5 years agoMerge pull request #15739 from dkurt:darknet_shortcut_asymm
Alexander Alekhin [Tue, 22 Oct 2019 12:52:13 +0000 (12:52 +0000)]
Merge pull request #15739 from dkurt:darknet_shortcut_asymm

5 years agoMerge pull request #15756 from dankamongmen:dankamongmen/stitching_surf_default
Alexander Alekhin [Tue, 22 Oct 2019 12:51:29 +0000 (12:51 +0000)]
Merge pull request #15756 from dankamongmen:dankamongmen/stitching_surf_default

5 years agostitching_detailed: use correct match_conf default
nick black [Tue, 22 Oct 2019 06:16:29 +0000 (02:16 -0400)]
stitching_detailed: use correct match_conf default

The usage function states that the default for match_conf is
0.65 if the default SURF feature finder is used, and 0.3 for
orbs. Indeed, if --feature orbs is used, match_conf is set
to 0.3f. This is a NOP, because the real default is also set
to 0.3f. Change it to 0.65f when SURF is in play.

5 years agoMerge pull request #15743 from collinbrake:feature_grammar_fixes_2
Alexander Alekhin [Sun, 20 Oct 2019 13:08:28 +0000 (13:08 +0000)]
Merge pull request #15743 from collinbrake:feature_grammar_fixes_2

5 years agoMerge pull request #14518 from SSteve:intersectConvexConvex-example
Alexander Alekhin [Sun, 20 Oct 2019 13:06:27 +0000 (13:06 +0000)]
Merge pull request #14518 from SSteve:intersectConvexConvex-example

5 years agoAdd documentation and example program for intersectConvexConvex
Steve Nicholson [Thu, 9 May 2019 00:54:17 +0000 (17:54 -0700)]
Add documentation and example program for intersectConvexConvex

5 years agoFix Darknet eltwise
Dmitry Kurtaev [Sat, 19 Oct 2019 09:54:15 +0000 (12:54 +0300)]
Fix Darknet eltwise

5 years agogrammar corrections for python gui docs
collin [Fri, 18 Oct 2019 21:42:56 +0000 (17:42 -0400)]
grammar corrections for python gui docs

5 years agoMerge pull request #15685 from pmur:cnz64f-simd
Alexander Alekhin [Fri, 18 Oct 2019 20:19:40 +0000 (20:19 +0000)]
Merge pull request #15685 from pmur:cnz64f-simd

5 years agoMerge pull request #15740 from xerus:fix_typo
Alexander Alekhin [Fri, 18 Oct 2019 18:41:35 +0000 (18:41 +0000)]
Merge pull request #15740 from xerus:fix_typo

5 years agoMerge pull request #15741 from mipsopen-fwu:issue_15730
Alexander Alekhin [Fri, 18 Oct 2019 18:40:25 +0000 (18:40 +0000)]
Merge pull request #15741 from mipsopen-fwu:issue_15730

5 years agoFix issue 15730.
Fei Wu [Fri, 18 Oct 2019 16:36:18 +0000 (00:36 +0800)]
Fix issue 15730.

5 years agoOpenCVFindMKL: Fix a typo
Pavel Grunt [Fri, 18 Oct 2019 16:04:48 +0000 (18:04 +0200)]
OpenCVFindMKL: Fix a typo

5 years agoEnable Eltwise layer with different numbers of inputs channels
Dmitry Kurtaev [Fri, 18 Oct 2019 15:51:52 +0000 (18:51 +0300)]
Enable Eltwise layer with different numbers of inputs channels

5 years agoMerge pull request #15737 from alalek:issue_15705
Alexander Alekhin [Fri, 18 Oct 2019 14:52:25 +0000 (14:52 +0000)]
Merge pull request #15737 from alalek:issue_15705

5 years agocore(simd): v_reverse() for MSA backend
Alexander Alekhin [Fri, 18 Oct 2019 13:05:03 +0000 (16:05 +0300)]
core(simd): v_reverse() for MSA backend

5 years agoMerge pull request #15716 from alalek:javadoc_fix
Alexander Alekhin [Thu, 17 Oct 2019 22:35:14 +0000 (22:35 +0000)]
Merge pull request #15716 from alalek:javadoc_fix

5 years agoMerge pull request #15180 from terfendail:wintr_stereobm
Alexander Alekhin [Thu, 17 Oct 2019 20:18:30 +0000 (20:18 +0000)]
Merge pull request #15180 from terfendail:wintr_stereobm

5 years agoMerge pull request #15709 from alalek:js_simd_reverse
Alexander Alekhin [Thu, 17 Oct 2019 13:14:50 +0000 (13:14 +0000)]
Merge pull request #15709 from alalek:js_simd_reverse

5 years agoMerge pull request #15708 from alalek:js_simd_support_1.38.48
Alexander Alekhin [Thu, 17 Oct 2019 13:14:34 +0000 (13:14 +0000)]
Merge pull request #15708 from alalek:js_simd_support_1.38.48

5 years agoMerge pull request #15718 from alalek:pylint_warnings
Alexander Alekhin [Thu, 17 Oct 2019 10:46:02 +0000 (10:46 +0000)]
Merge pull request #15718 from alalek:pylint_warnings

5 years agoMerge pull request #15720 from alalek:cmake_fix_uwp
Alexander Smorkalov [Thu, 17 Oct 2019 10:35:27 +0000 (10:35 +0000)]
Merge pull request #15720 from alalek:cmake_fix_uwp

5 years agoMerge pull request #15722 from jasjuang:3.4
Alexander Smorkalov [Thu, 17 Oct 2019 10:19:47 +0000 (10:19 +0000)]
Merge pull request #15722 from jasjuang:3.4

5 years agodocument CC_STAT_MAX in ConnectedComponentsTypes
jasjuang [Thu, 17 Oct 2019 00:22:25 +0000 (17:22 -0700)]
document CC_STAT_MAX in ConnectedComponentsTypes

5 years agocmake: fix UWP scripts path
Alexander Alekhin [Wed, 16 Oct 2019 21:42:31 +0000 (21:42 +0000)]
cmake: fix UWP scripts path

5 years agoMerge pull request #15715 from adamrankin:patch-1
Alexander Alekhin [Wed, 16 Oct 2019 18:52:10 +0000 (18:52 +0000)]
Merge pull request #15715 from adamrankin:patch-1

5 years agofix pylint warnings
Alexander Alekhin [Wed, 16 Oct 2019 15:49:33 +0000 (18:49 +0300)]
fix pylint warnings

pylint 1.8.3

5 years agojavadoc: fix generation with OpenJDK 11
Alexander Alekhin [Wed, 16 Oct 2019 14:42:20 +0000 (17:42 +0300)]
javadoc: fix generation with OpenJDK 11

5 years agoCOMP: Enabling build with recent VTK version
Adam Rankin [Wed, 16 Oct 2019 14:04:41 +0000 (10:04 -0400)]
COMP: Enabling build with recent VTK version

VTK_MAJOR_VERSION not found unless header is included

5 years agoMerge pull request #15701 from alalek:issue_15691
Alexander Alekhin [Wed, 16 Oct 2019 11:13:07 +0000 (11:13 +0000)]
Merge pull request #15701 from alalek:issue_15691

5 years agoMerge pull request #15700 from alalek:issue_12943
Alexander Alekhin [Wed, 16 Oct 2019 11:12:49 +0000 (11:12 +0000)]
Merge pull request #15700 from alalek:issue_12943

5 years agojs(simd): v_reverse implementation
Alexander Alekhin [Tue, 15 Oct 2019 15:46:08 +0000 (18:46 +0300)]
js(simd): v_reverse implementation

5 years agojs(simd): support Emscripten 1.38.48-upstream
Alexander Alekhin [Tue, 15 Oct 2019 12:29:10 +0000 (15:29 +0300)]
js(simd): support Emscripten 1.38.48-upstream

5 years agocore(alloc): force initialization of memalign flag
Alexander Alekhin [Mon, 14 Oct 2019 15:39:54 +0000 (18:39 +0300)]
core(alloc): force initialization of memalign flag

- before main() launch

5 years agocore(ipp): skip huge input in flip()
Alexander Alekhin [Mon, 14 Oct 2019 15:10:32 +0000 (18:10 +0300)]
core(ipp): skip huge input in flip()

- IPP/SSE4.2 works well

5 years agoMerge pull request #15688 from JamesNewton:3.4
Alexander Smorkalov [Sat, 12 Oct 2019 18:05:54 +0000 (18:05 +0000)]
Merge pull request #15688 from JamesNewton:3.4

5 years agoUpdate js_image_arithmetics.markdown
JamesNewton [Fri, 11 Oct 2019 17:07:22 +0000 (10:07 -0700)]
Update js_image_arithmetics.markdown

5 years agoMerge pull request #15662 from ChipKerchner:addVReverseIntrinsic
Chip Kerchner [Fri, 11 Oct 2019 15:34:17 +0000 (11:34 -0400)]
Merge pull request #15662 from ChipKerchner:addVReverseIntrinsic

* New v_reverse HAL intrinsic for reversing the ordering of a vector

* Fix conflict.

* Try to resolve conflict again.

* Try one more time.

* Add _MM_SHUFFLE. Remove non-vectorize code in SSE2. Fix copy and paste issue with NEON.

* Change v_uint16x8 SSE2 version to use shuffles

5 years agoMerge pull request #15527 from everton1984:faster_acc
Everton Constantino [Fri, 11 Oct 2019 15:32:59 +0000 (12:32 -0300)]
Merge pull request #15527 from everton1984:faster_acc

* Adding support for vectorized masking for uchar/ushort.

* Fixing bug where mask was zeroing the dst. Improved the way to calculate
the mask and tweaked for further performance improvements.

* Fixing mask comparison test.

* Restricting to one channel.

* Adding support for 3 channels, switch old approach to start using HAL's
v_select.

5 years agocore: vectorize countNonZero64f
Paul E. Murphy [Wed, 28 Aug 2019 17:09:07 +0000 (12:09 -0500)]
core: vectorize countNonZero64f

Improves performance a bit. 2.2x on P9 and 2 - 3x on coffee lake
x86-64.

5 years agoMerge tag '3.4.8'
Alexander Alekhin [Wed, 9 Oct 2019 15:33:35 +0000 (18:33 +0300)]
Merge tag '3.4.8'

5 years agoOpenCV release (3.4.8)
Alexander Alekhin [Wed, 9 Oct 2019 11:42:29 +0000 (14:42 +0300)]
OpenCV release (3.4.8)

OpenCV 3.4.8

5 years agoMerge pull request #15544 from mshabunin:disable_posix_memalign
Maksim Shabunin [Wed, 9 Oct 2019 11:06:12 +0000 (14:06 +0300)]
Merge pull request #15544 from mshabunin:disable_posix_memalign

* Disable posix_memalign by default

* core: fix memalign parameter handling

5 years agoMerge pull request #15666 from seanm:Wnewline
Alexander Smorkalov [Wed, 9 Oct 2019 11:04:44 +0000 (11:04 +0000)]
Merge pull request #15666 from seanm:Wnewline

5 years agoMerge pull request #15658 from tolysz:patch-1
Marcin Tolysz [Wed, 9 Oct 2019 08:38:10 +0000 (09:38 +0100)]
Merge pull request #15658 from tolysz:patch-1

* Cuda + OpenGL on ARM

There might be multiple ways of getting OpenCV compile on Tegra (NVIDIA Jetson) platform, but mainly they modify CUDA(8,9,10...) source code, this one fixes it for all installations.
( https://devtalk.nvidia.com/default/topic/1007290/jetson-tx2/building-opencv-with-opengl-support-/post/5141945/#5141945 et al.).
This way is exactly the same as the one proposed but the code change happens in OpenCV.

* Updated,
The link provided mentions: cuda8 + 9, I have cuda 10 + 10.1 (and can confirm it is still defined this way).
NVIDIA is probably using some other "secret" backend with Jetson.

5 years agoFixed clang -Wnewline-eof warning by adding newline to end of file
Sean McBride [Tue, 8 Oct 2019 18:16:11 +0000 (14:16 -0400)]
Fixed clang -Wnewline-eof warning by adding newline to end of file

5 years agoMerge pull request #15661 from alalek:fix_android_build_avx2
Alexander Alekhin [Tue, 8 Oct 2019 23:36:26 +0000 (23:36 +0000)]
Merge pull request #15661 from alalek:fix_android_build_avx2

5 years agoMerge pull request #15664 from alalek:build_eliminate_cuda_warnings
Alexander Alekhin [Tue, 8 Oct 2019 15:27:54 +0000 (15:27 +0000)]
Merge pull request #15664 from alalek:build_eliminate_cuda_warnings

5 years agobuild: eliminate CUDA warnings
Alexander Alekhin [Tue, 8 Oct 2019 12:30:02 +0000 (15:30 +0300)]
build: eliminate CUDA warnings

5 years agobuild: update AVX2 check
Alexander Alekhin [Tue, 8 Oct 2019 10:12:02 +0000 (13:12 +0300)]
build: update AVX2 check

- _mm256_bslli_epi128() works in GCC 4.9.3+ only
- Android NDK r10 doesn't support this instruction

5 years agoMerge pull request #15642 from alalek:issue_15597
Alexander Alekhin [Mon, 7 Oct 2019 21:33:20 +0000 (00:33 +0300)]
Merge pull request #15642 from alalek:issue_15597

5 years agoMerge pull request #15510 from seiko2plus:issue15506
Sayed Adel [Mon, 7 Oct 2019 19:01:35 +0000 (21:01 +0200)]
Merge pull request #15510 from seiko2plus:issue15506

* core: rework and optimize SIMD implementation of dotProd

  - add new universal intrinsics v_dotprod[int32], v_dotprod_expand[u&int8, u&int16, int32], v_cvt_f64(int64)
  - add a boolean param for all v_dotprod&_expand intrinsics that change the behavior of addition order between
    pairs in some platforms in order to reach the maximum optimization when the sum among all lanes is what only matters
  - fix clang build on ppc64le
  - support wide universal intrinsics for dotProd_32s
  - remove raw SIMD and activate universal intrinsics for dotProd_8
  - implement SIMD optimization for dotProd_s16&u16
  - extend performance test data types of dotprod
  - fix GCC VSX workaround of vec_mule and vec_mulo (in little-endian it must be swapped)
  - optimize v_mul_expand(int32) on VSX

* core: remove boolean param from v_dotprod&_expand and implement v_dotprod_fast&v_dotprod_expand_fast

  this changes made depend on "terfendail" review

5 years agoMerge pull request #15654 from sturkmen72:patch-3
Alexander Alekhin [Mon, 7 Oct 2019 16:15:04 +0000 (16:15 +0000)]
Merge pull request #15654 from sturkmen72:patch-3

5 years agoimgproc: update histogram test
Alexander Alekhin [Mon, 7 Oct 2019 12:06:43 +0000 (15:06 +0300)]
imgproc: update histogram test

5 years agoMerge pull request #15653 from tolysz:patch-1
Alexander Alekhin [Mon, 7 Oct 2019 12:04:13 +0000 (12:04 +0000)]
Merge pull request #15653 from tolysz:patch-1

5 years agoFix compiler warnings for latest cuda npp which defines this itself as:
Marcin Tolysz [Mon, 7 Oct 2019 10:45:26 +0000 (11:45 +0100)]
Fix compiler warnings for latest cuda npp which defines this itself as:
```
#define NPP_VER_MAJOR 10
#define NPP_VER_MINOR 2
#define NPP_VER_PATCH 0
#define NPP_VER_BUILD 243

#define NPP_VERSION (NPP_VER_MAJOR * 1000 +     \
                     NPP_VER_MINOR *  100 +     \
                     NPP_VER_PATCH)

5 years agoUpdate copy.cpp
Suleyman TURKMEN [Mon, 7 Oct 2019 08:59:52 +0000 (11:59 +0300)]
Update copy.cpp

5 years agoMerge pull request #15646 from alalek:fix_avx512_detection
Alexander Alekhin [Sat, 5 Oct 2019 15:30:09 +0000 (15:30 +0000)]
Merge pull request #15646 from alalek:fix_avx512_detection

5 years agoMerge pull request #15644 from wise86-android:upgradePySetup
Alexander Alekhin [Sat, 5 Oct 2019 15:29:13 +0000 (15:29 +0000)]
Merge pull request #15644 from wise86-android:upgradePySetup

5 years agoMerge pull request #15641 from alalek:dnn_pytest_update_3.4
Alexander Alekhin [Sat, 5 Oct 2019 15:28:52 +0000 (15:28 +0000)]
Merge pull request #15641 from alalek:dnn_pytest_update_3.4

5 years agoMerge pull request #15602 from alalek:core_softfloat_ubsan_shift
Alexander Alekhin [Sat, 5 Oct 2019 15:27:35 +0000 (15:27 +0000)]
Merge pull request #15602 from alalek:core_softfloat_ubsan_shift

5 years agoupgrade the file to use the correct package name and include py3/gtk3 support
Wise86 [Sat, 5 Oct 2019 09:48:09 +0000 (11:48 +0200)]
upgrade the file to use the correct package name and include py3/gtk3 support

5 years agofix avx512 detection
Alexander Alekhin [Sat, 5 Oct 2019 10:39:35 +0000 (10:39 +0000)]
fix avx512 detection

- renamed Cascade Lake AVX512_CEL => AVX512_CLX (align with Intel SDE tool)
- fixed CLX instruction sets (no IFMA/VBMI)
- added flag to bypass CPU baseline check: OPENCV_SKIP_CPU_BASELINE_CHECK

5 years agoMerge pull request #15626 from alalek:dnn_openvino_2019r3
Alexander Alekhin [Fri, 4 Oct 2019 19:45:37 +0000 (19:45 +0000)]
Merge pull request #15626 from alalek:dnn_openvino_2019r3

5 years agoimgproc: accurate histogram value thresholding
Alexander Alekhin [Thu, 3 Oct 2019 13:02:17 +0000 (16:02 +0300)]
imgproc: accurate histogram value thresholding

5 years agoMerge pull request #15622 from dkurt:enet_ie_cpu
Alexander Alekhin [Fri, 4 Oct 2019 16:31:05 +0000 (16:31 +0000)]
Merge pull request #15622 from dkurt:enet_ie_cpu

5 years agodnn(pytest): backport changes from master, update test setUp()
Alexander Alekhin [Fri, 4 Oct 2019 15:35:11 +0000 (18:35 +0300)]
dnn(pytest): backport changes from master, update test setUp()

5 years agoEnable ENet with Inference Engine backend on CPU
Dmitry Kurtaev [Wed, 2 Oct 2019 11:05:41 +0000 (14:05 +0300)]
Enable ENet with Inference Engine backend on CPU

5 years agodnn: update IE tests
Alexander Alekhin [Fri, 4 Oct 2019 07:29:27 +0000 (07:29 +0000)]
dnn: update IE tests

5 years agoMerge pull request #15634 from alalek:issue_15083_4903
Alexander Alekhin [Fri, 4 Oct 2019 11:54:54 +0000 (11:54 +0000)]
Merge pull request #15634 from alalek:issue_15083_4903

5 years agoMerge pull request #15637 from APrigarina:3.4_version2
Alexander Alekhin [Fri, 4 Oct 2019 11:09:35 +0000 (14:09 +0300)]
Merge pull request #15637 from APrigarina:3.4_version2

QR-Code detection: accuracy improvement (fixed)

5 years agobug fix
APrigarina [Fri, 4 Oct 2019 08:47:56 +0000 (11:47 +0300)]
bug fix

5 years agoimgproc: fix fitLine() implementation
Alexander Alekhin [Thu, 3 Oct 2019 20:45:43 +0000 (20:45 +0000)]
imgproc: fix fitLine() implementation

- update optimal solutions on each iteration

5 years agodnn: use OpenVINO 2019R3 defines
Alexander Alekhin [Mon, 2 Sep 2019 12:35:35 +0000 (15:35 +0300)]
dnn: use OpenVINO 2019R3 defines

5 years agoMerge pull request #15621 from alalek:cmake_ipp_licenses
Alexander Alekhin [Wed, 2 Oct 2019 14:30:17 +0000 (14:30 +0000)]
Merge pull request #15621 from alalek:cmake_ipp_licenses

5 years agoMerge pull request #15620 from alalek:issue_14727
Alexander Alekhin [Wed, 2 Oct 2019 14:29:55 +0000 (14:29 +0000)]
Merge pull request #15620 from alalek:issue_14727

5 years agoMerge pull request #15467 from elmsfu:dshow/add_save_graph_file
Alexander Alekhin [Wed, 2 Oct 2019 14:27:56 +0000 (14:27 +0000)]
Merge pull request #15467 from elmsfu:dshow/add_save_graph_file

5 years agoMerge pull request #15619 from alalek:dnn_eltwise_sum_ie_ocl
Alexander Alekhin [Tue, 1 Oct 2019 18:02:45 +0000 (18:02 +0000)]
Merge pull request #15619 from alalek:dnn_eltwise_sum_ie_ocl

5 years agodshow: Add ability to save direct show graph
Elms [Thu, 5 Sep 2019 20:51:37 +0000 (13:51 -0700)]
dshow: Add ability to save direct show graph

5 years agodnn(test): skip IE/OCL test for "sum"
Alexander Alekhin [Tue, 1 Oct 2019 15:29:47 +0000 (18:29 +0300)]
dnn(test): skip IE/OCL test for "sum"

5 years agocmake: add check for IPP IW license files
Alexander Alekhin [Tue, 1 Oct 2019 15:04:21 +0000 (18:04 +0300)]
cmake: add check for IPP IW license files

5 years agoMerge pull request #15618 from alalek:dnn_test_async_timeout
Alexander Alekhin [Tue, 1 Oct 2019 14:48:47 +0000 (14:48 +0000)]
Merge pull request #15618 from alalek:dnn_test_async_timeout

5 years agocore: workaround transform() inplace calls
Alexander Alekhin [Tue, 1 Oct 2019 13:52:14 +0000 (16:52 +0300)]
core: workaround transform() inplace calls

5 years agodnn: increase async test timeout
Alexander Alekhin [Tue, 1 Oct 2019 10:31:57 +0000 (13:31 +0300)]
dnn: increase async test timeout

5 years agoMerge pull request #15609 from sturkmen72:update_loadsave_cpp
Alexander Alekhin [Mon, 30 Sep 2019 18:37:21 +0000 (18:37 +0000)]
Merge pull request #15609 from sturkmen72:update_loadsave_cpp

5 years agoMerge pull request #15480 from cancerberoSgx:estimateAffine2D
Sebastián Gurin [Mon, 30 Sep 2019 18:35:26 +0000 (15:35 -0300)]
Merge pull request #15480 from cancerberoSgx:estimateAffine2D

js - cv.estimateAffine2D, cv.warpPolar

5 years agoMerge pull request #15356 from APrigarina:3.4_version2
ann [Mon, 30 Sep 2019 18:33:58 +0000 (21:33 +0300)]
Merge pull request #15356 from APrigarina:3.4_version2

QR-Code detection: accuracy improvement

* first commit

* resize improvement

* resize improvement

* dummy commit

* warnings fix

* warnings fix

* warnings fix

* test added

* resize fix

* resize fix

* tests changed

* tests changed

5 years agoUpdate loadsave.cpp
Suleyman TURKMEN [Sat, 28 Sep 2019 20:24:17 +0000 (23:24 +0300)]
Update loadsave.cpp

5 years agocore: fix UBSAN in softfloat
Alexander Alekhin [Thu, 26 Sep 2019 17:00:30 +0000 (20:00 +0300)]
core: fix UBSAN in softfloat

5 years agoMerge pull request #15600 from alalek:imgproc_pyramid_fix_indexing
Alexander Alekhin [Fri, 27 Sep 2019 10:21:52 +0000 (10:21 +0000)]
Merge pull request #15600 from alalek:imgproc_pyramid_fix_indexing

5 years agoMerge pull request #15595 from frastlin:rodrigues
Brandon [Fri, 27 Sep 2019 10:06:07 +0000 (03:06 -0700)]
Merge pull request #15595 from frastlin:rodrigues

* Added Rodrigues to the JS embindgen.py and a unittest in test_calib3d.js on 3.4 base

5 years agoimgproc: fix indexing issue in pyramids
Alexander Alekhin [Thu, 26 Sep 2019 14:56:24 +0000 (17:56 +0300)]
imgproc: fix indexing issue in pyramids

UBSAN violation expression: 'tab = tabR - x;'

5 years agoMerge pull request #15580 from smbz:dnn-lstm-reverse
Alexander Alekhin [Wed, 25 Sep 2019 15:54:05 +0000 (15:54 +0000)]
Merge pull request #15580 from smbz:dnn-lstm-reverse

5 years agoMerge pull request #15579 from alalek:ocl_use_host_mem_ptr_flag
Alexander Alekhin [Wed, 25 Sep 2019 15:12:58 +0000 (15:12 +0000)]
Merge pull request #15579 from alalek:ocl_use_host_mem_ptr_flag

5 years agodnn: Allow LSTM layer to operate in reverse direction
Andrew Ryrie [Wed, 25 Sep 2019 13:12:43 +0000 (14:12 +0100)]
dnn: Allow LSTM layer to operate in reverse direction

This is useful for bidirectional LSTMs.

5 years agoMerge pull request #15582 from terfendail:pyrdown_oob
Alexander Alekhin [Wed, 25 Sep 2019 12:37:31 +0000 (12:37 +0000)]
Merge pull request #15582 from terfendail:pyrdown_oob

5 years agoMerge pull request #15572 from l-bat:deconv3d
Lubov Batanina [Wed, 25 Sep 2019 12:35:04 +0000 (15:35 +0300)]
Merge pull request #15572 from l-bat:deconv3d

Fix computation of internal shapes in Deconvolution layer

* Fix computation of internal shapes

* Refactoring

5 years agocore(ocl): add flag OPENCV_OPENCL_ENABLE_MEM_USE_HOST_PTR
Alexander Alekhin [Tue, 24 Sep 2019 10:03:29 +0000 (13:03 +0300)]
core(ocl): add flag OPENCV_OPENCL_ENABLE_MEM_USE_HOST_PTR

to control CL_MEM_USE_HOST_PTR usage

5 years agoStereoBM algorithm updated to use wide universal intrinsics
Vitaly Tuzov [Mon, 1 Jul 2019 14:44:49 +0000 (17:44 +0300)]
StereoBM algorithm updated to use wide universal intrinsics

5 years agoMerge pull request #15581 from takacsd:3.4
Alexander Alekhin [Wed, 25 Sep 2019 10:45:49 +0000 (10:45 +0000)]
Merge pull request #15581 from takacsd:3.4