* Fix wrong MSAN errors.
Because Fortran is called in Lapack, MSAN does not think the memory
has been written even though it is the case.
MSAN does no support well cross-language memory analysis.
* Make a dedicated check.
- Add special case handling when submodule has the same name as parent
- `PyDict_SetItemString` doesn't steal reference, so reference count
should be explicitly decremented to transfer object life-time
ownership
- Add sanity checks for module registration input
- Add Python 2 and Python 3 reference counting handling
G-API: Wrap GStreamerSource
* Wrap GStreamerSource into python
* Fixed test skipping when can't make Gst-src
* Wrapped GStreamerPipeline class, added dummy test for it
* Fix no_gst testing
* Changed wrap for GStreamerPipeline::getStreamingSource() : now python-specific in-class method GStreamerPipeline::get_streaming_source()
* Added accuracy tests vs OCV:VideoCapture(Gstreamer)
* Add skipping when can't use VideoCapture(GSTREAMER);
Add better handling of GStreamer backend unavailable;
Changed video to avoid terminations
* Applying comments
* back to a separate get_streaming_source function, with comment
Co-authored-by: OrestChura <orest.chura@intel.com>
G-API: oneVPL DX11 inference
* Draft GPU infer
* Fix incorrect subresource_id for array of textures
* Fix for TheOneSurface in different Frames
* Turn on VPP param configuration
* Add cropIn params
* Remove infer sync sample
* Remove comments
* Remove DX11AllocResource extra init
* Add condition for NV12 processing in giebackend
* Add VPP frames pool param configurable
* -M Remove extra WARN & INFOs, Fix custom MAC
* Remove global vars from example, Fix some comments, Disable blobParam due to OV issue
* Conflict resolving
* Revert back pointer cast for cv::any
clang-cl defines both __clang__ and _MSC_VER, yet uses `#pragma GCC` to disable certain diagnostics.
At the time `-Wreturn-type-c-linkage` was reported by clang-cl.
This PR fixes this behavior by reordering defines.
- Add special case handling when submodule has the same name as parent
- `PyDict_SetItemString` doesn't steal reference, so reference count
should be explicitly decremented to transfer object life-time
ownership
- Add sanity checks for module registration input
Comment from Python documentation:
Unlike other functions that steal references, `PyModule_AddObject()` only
decrements the reference count of value on success.
This means that its return value must be checked, and calling code must
`Py_DECREF()` value manually on error.
GAPI: Add OAK backend
* Initial tests and cmake integration
* Add a public header and change tests
* Stub initial empty template for the OAK backend
* WIP
* WIP
* WIP
* WIP
* Runtime dai hang debug
* Refactoring
* Fix hang and debug frame data
* Fix frame size
* Fix data size issue
* Move test code to sample
* tmp refactoring
* WIP: Code refactoring except for the backend
* WIP: Add non-camera sample
* Fix samples
* Backend refactoring wip
* Backend rework wip
* Backend rework wip
* Remove mat encoder
* Fix namespace
* Minor backend fixes
* Fix hetero sample and refactor backend
* Change linking logic in the backend
* Fix oak sample
* Fix working with ins/outs in OAK island
* Trying to fix nv12 problem
* Make both samples work
* Small refactoring
* Remove meta args
* WIP refactoring kernel API
* Change in/out args API for kernels
* Fix build
* Fix cmake warning
* Partially address review comments
* Partially address review comments
* Address remaining comments
* Add memory ownership
* Change pointer-to-pointer to reference-to-pointer
* Remove unnecessary reference wrappers
* Apply review comments
* Check that graph contains only one OAK island
* Minor refactoring
* Address review comments
Add support for 3D point cloud segmentation, using the USAC framework.
* Modify the RANSAC framework in usac such that RANSAC can be used in 3D point cloud segmentation.
* 1. Add support for 3D point cloud segmentation, using the USAC framework.
2. Add solvers, error estimators for plane model and sphere model.
* Added code samples to the comments of class SACSegmentation.
* 1. Update the segment interface parameters of SACSegmentation.
2. Fix some errors in variable naming.
* Add tests for plane detection.
* 1. Add tests for sphere segmentation.
2. Fix some bugs found by tests.
3. Rename "segmentation" to "sac segmentation".
4. Rename "detect" to "segment".
TODO: Too much duplicate code, the structure of the test needs to be rebuilt.
* 1. Use SIMD acceleration for plane model and sphere model error estimation.
2. Optimize the RansacQualityImpl#getScore function to avoid multiple calls to the error#getError function.
3. Fix a warning in test_sac_segmentation.cpp.
* 1. Fix the warning of ModelConstraintFunction ambiguity.
2. Fix warning: no previous declaration for'void cv::usac::modelParamsToUsacConfig(cv::Ptr<cv::usac::SimpleUsacConfig>&, const cv::Ptr<const cv::usac::Model>& )
* Fix a warning in test_sac_segmentation.cpp about direct comparison of different types of data.
* Add code comments related to the interpretation of model coefficients.
* Update the use of custom model constraint functions.
* Simplified test code structure.
* Update the method of checking plane models.
* Delete test for cylinder.
* Add some comments about UniversalRANSAC.
* 1. The RANSAC paper in the code comments is referenced using the bibtex format.
2. The sample code in the code comments is replaced using @snippet.
3. Change the public API class SACSegmentation to interface.
4. Clean up the old useless code.
* fix warning(no previous declaration) in 3d_sac_segmentation.cpp.
* Fix compilation errors caused by 3d_sac_segmentation.cpp.
* Move the function sacModelMinimumSampleSize() from ptcloud.hpp to sac_segmentation.cpp.
* 1. Change the interface for setting the number of threads to the interface for setting whether to be parallel.
2. Move interface implementation code in ptcloud_utils.hpp to ptcloud_utils.cpp.
* SACSegmentation no longer inherits Algorithm.
* Add the constructor and destructor of SACSegmentation.
* 1. For the declaration of the common API, the prefix and suffix of the parameter names no longer contain underscores.
2. Rename the function _getMatFromInputArray -> getPointsMatFromInputArray.
3. Change part of CV_CheckDepth to CV_CheckDepthEQ.
4. Remove the doxygen flag from the source code.
5. Update the loop termination condition of SIMD in the point cloud section of 3D module.
* fix warning: passing 'bool' chooses 'int' over 'size_t {aka unsigned int}' .
* fix warning: passing 'bool' chooses 'int' over 'size_t {aka unsigned int}' .
The Qt backend directly calls some OpenGL functions (glClear, glHint,
glViewport), but since OCV 4.5.5 the GL libraries are no longer part
of the global extra dependencies. When linking with "-Wl,--no-undefined"
this causes linker errors:
`opencv-4.5.5/modules/highgui/src/window_QT.cpp:3307: undefined reference to `glClear'`
Fixes: #21346
Related issues: #21299
* Fix compile against lapack-3.10.0
Fix compilation against lapack >= 3.9.1 and 3.10.0 while not breaking older versions
OpenCVFindLAPACK.cmake & CMakeLists.txt: determine OPENCV_USE_LAPACK_PREFIX from LAPACK_VERSION
hal_internal.cpp : Only apply LAPACK_FUNC to functions whose number of inputs depends on LAPACK_FORTRAN_STR_LEN in lapack >= 3.9.1
lapack_check.cpp : remove LAPACK_FUNC which is not OK as function are not used with input parameters (so lapack.h preprocessing of "LAPACK_xxxx(...)" is not applicable with lapack >= 3.9.1
If not removed lapack_check fails so LAPACK is deactivated in build (not want we want)
use OCV_ prefix and don't use Global, instead generate OCV_LAPACK_FUNC depending on CMake Conditions
Remove CONFIG from find_package(LAPACK) and use LAPACK_GLOBAL and LAPACK_NAME to figure out if using netlib's reference LAPACK implementation and how to #define OCV_LAPACK_FUNC(f)
* Fix typos and grammar in comments
Fixed threshold(THRESH_TOZERO) at imgproc(IPP)
* Fixed#16085: imgproc(IPP): wrong result from threshold(THRESH_TOZERO)
* 1. Added test cases with float where all bits of mantissa equal 1, min and max float as inputs
2. Used nextafterf instead of cast to hex
* Used float value in test instead of hex and casts
* Changed input value in test
When computing:
t1 = (bayer[1] + bayer[bayer_step] + bayer[bayer_step+2] + bayer[bayer_step*2+1])*G2Y;
there is a T (unsigned short or char) multiplied by an int which can overflow.
Then again, it is stored to t1 which is unsigned so the overflow disappears.
Keeping all unsigned is safer.
Further optimize DNN for RISC-V Vector.
* Optimize DNN on RVV by using vsetvl.
* Rename vl.
* Update fastConv by using setvl instead of mask.
* Fix fastDepthwiseConv
- QGLWidget changed to QOpenGLWidget in window_QT.h for Qt6 using
typedef OpenCVQtWidgetBase for handling Qt version
- Implement Qt6/OpenGL functionality in window_QT.cpp
- Swap QGLWidget:: function calls for OpenCVQtWidgetBase:: function calls
- QGLWidget::updateGL deprecated, swap to QOpenGLWidget::update for Qt6
- Add preprocessor definition to detect Qt6 -- HAVE_QT6
- Add OpenGLWidgets to qdeps list in highgui CMakeLists.txt
- find_package CMake command added for locating Qt module OpenGLWidgets
- Added check that Qt6::OpenGLWidgets component is found. Shut off Qt-openGL functionality if not found.
Fow now, it is possible to define valid rectangle for which some
functions overflow (e.g. br(), ares() ...).
This patch fixes the intersection operator so that it works with
any rectangle.
G-API: oneVPL merge DX11 acceleration
* Merge DX11 initial
* Fold conditions row in MACRO in utils
* Inject DeviceSelector
* Turn on DeviceSelector in DX11
* Change sharedLock logic & Move FMT checking in FrameAdapter c-tor
* Move out NumSuggestFrame to configure params
* Drain file source fix
* Fix compilation
* Force zero initializetion of SharedLock
* Fix some compiler warnings
* Fix integer comparison warnings
* Fix integers in sample
* Integrate Demux
* Fix compilation
* Add predefined names for some CfgParam
* Trigger CI
* Fix MultithreadCtx bug, Add Dx11 GetBlobParam(), Get rif of ATL CComPtr
* Fix UT: remove unit test with deprecated video from opencv_extra
* Add creators for most usable CfgParam
* Eliminate some warnings
* Fix warning in GAPI_Assert
* Apply comments
* Add VPL wrapped header with MSVC pragma to get rid of global warning masking
Added CV_PROP_RW macro to keypoints
* Added CV_PROP_RW macro to keypoints
As outlined in the feature request in the issue https://github.com/opencv/opencv/issues/21171 : the keypoints field has been made parsable by the bindings.
* Added test for keypoints
Added test to check if the CV_PROP_RW macro added in the previous commit makes keypoints public and accessible through the python API.
Audio MSMF: added the ability to set sample per second
* Audio MSMF: added the ability to set sample per second
* changed the valid sampling rate check
* fixed docs
* add test
* fixed warning
* fixed error
* fixed error
Update RVV backend for using Clang.
* Update cmake file of clang.
* Modify the RVV optimization on DNN to adapt to clang.
* Modify intrin_rvv: Disable some existing types.
* Modify intrin_rvv: Reinterpret instead of load&cast.
* Modify intrin_rvv: Update load&store without cast.
* Modify intrin_rvv: Rename vfredsum to fredosum.
* Modify intrin_rvv: Rewrite Check all/any by using vpopc.
* Modify intrin_rvv: Use reinterpret instead of c-style casting.
* Remove all macros which is not used in v_reinterpret
* Rename vpopc to vcpop according to spec.
New odometry Pipeline
* first intergation
* tests run, but not pass
* add previous version of sigma calc
* add minor comment
* strange fixes
* fix fast ICP
* test changes; fast icp still not work correctly
* finaly, it works
* algtype fix
* change affine comparison
* boolean return
* fix bug with angle and cos
* test pass correctly
* fix for kinfu pipeline
* add compute points normals
* update for new odometry
* change odometry_evaluation
* odometry_evaluation works
* change debug logs
* minor changes
* change depth setting in odometryFrame
* fastICP works with 4num points
* all odometries work with 4mun points
* odometry full works on 4num points and normals
* replace ICP with DEPTH; comments replacements
* create prepareFrame; add docs for Odometry
* change getPyramids()
* delete extra code
* add intrinsics; but dont works
* bugfix with nan checking
* add gpu impl
* change createOdometryFrame func
* remove old fastICP code
* comments fix
* add comments
* minor fixes
* other minor fixes
* add channels assert
* add impl for odometry settings
* add pimpl to odometry
* linux warning fix
* linux warning fix 1
* linux warning fix 2
* linux error fix
* linux warning fix 3
* linux warning fix 4
* linux error fix 2
* fix test warnings
* python build fix
* doxygen fix
* docs fix
* change normal tests for 4channel point
* all Normal tests pass
* plane works
* add warp frame body
* minor fix
* warning fixes
* try to fix
* try to fix 1
* review fix
* lvls fix
* createOdometryFrame fix
* add comment
* const reference
* OPENCV_3D_ prefix
* const methods
* OdometryFramePyramidType ifx
* add assert
* precomp moved upper
* delete types_c
* add assert for get and set functions
* minor fixes
* remove core.hpp from header
* ocl_run add
* warning fix
* delete extra comment
* minor fix
* setDepth fix
* delete underscore
* odometry settings fix
* show debug image fix
* build error fix
* other minor fix
* add const to signatures
* fix
* conflict fix
* getter fix
* Fix integer overflow in cv::Luv2RGBinteger::process.
For LL=49, uu=205, vv=23, we end up with x=7373056 and y=458
which overflows y*x.
* imgproc(test): adjust test parameters to cover SIMD code
Accelerated 3D point cloud Farthest Point Sampling calculation using SIMD.
* Add several 3D point cloud sampling functions: Random, VoxelGrid, FarthestPoint.
* Made some code detail changes and exposed the random number generator parameters at the interface.
* Add simple tests for sampling.
* Modify interface output parameters.
* Modify interface return value.
* The sampling test is modified for the new changes of function interface.
* Improved test of VoxelGridFilterSampling
* Improved test of VoxelGridFilterSampling and FPS.
* Add test for the dist_lower_limit arguments of FPS function.
* Optimization function _getMatFromInputArray.
* Optimize the code style and some details according to the suggestions.
* Clear prefix cv: in the source code.
* Change the initialization of Mat in the sampling test.
* 1. Unified code style
2. Optimize randomSampling method
* 1. Optimize code comments.
2. Remove unused local variables.
* Rebuild the structure of the test, make the test case more reliable, and change the code style.
* Update test_sampling.cpp
Fix a warning.
* Use SIMD to optimize the farthest point sampling.
* Optimize the farthest point sampling SIMD code.
* 1. remove `\n` from the ptcloud.hpp comment.
2. updated the default value of the argument arrangement_of_points in the _getMatFromInputArray function in ptcloud_utils.hpp from 0 to 1, since the latter is more commonly used (such arrangement is easier for SIMD acceleration).
3. removed two functions in ptcloud_utils.hpp that were not used.
* Remove the <br> in the comment.
* Fix whitespace issues.
* dnn: LSTM optimisation
This uses the AVX-optimised fastGEMM1T for matrix multiplications where available, instead of the standard cv::gemm.
fastGEMM1T is already used by the fully-connected layer. This commit involves two minor modifications:
- Use unaligned access. I don't believe this involves any performance hit in on modern CPUs (Nehalem and Bulldozer onwards) in the case where the address is actually aligned.
- Allow for weight matrices where the number of columns is not a multiple of 8.
I have not enabled AVX-512 as I don't have an AVX-512 CPU to test on.
* Fix warning about initialisation order
* Remove C++11 syntax
* Fix build when AVX(2) is not available
In this case the CV_TRY_X macros are defined to 0, rather than being undefined.
* Minor changes as requested:
- Don't check hardware support for AVX(2) when dispatch is disabled for these
- Add braces
* Fix out-of-bounds access in fully connected layer
The old tail handling in fastGEMM1T implicitly rounded vecsize up to the next multiple of 8, and the fully connected layer implements padding up to the next multiple of 8 to cope with this. The new tail handling does not round the vecsize upwards like this but it does require that the vecsize is at least 8. To adapt to the new tail handling, the fully connected layer now rounds vecsize itself at the same time as adding the padding(which makes more sense anyway).
This also means that the fully connected layer always passes a vecsize of at least 8 to fastGEMM1T, which fixes the out-of-bounds access problems.
* Improve tail mask handling
- Use static array for generating tail masks (as requested)
- Apply tail mask to the weights as well as the input vectors to prevent spurious propagation of NaNs/Infs
* Revert whitespace change
* Improve readability of conditions for using AVX
* dnn(lstm): minor coding style changes, replaced left aligned load
