- 02 Jul, 2019 13 commits
-
-
Alexander Alekhin authored
-
Alexander Alekhin authored
-
Alexander Alekhin authored
-
Alexander Alekhin authored
-
Rafa Gomez-Jordana authored
-
Alexander Alekhin authored
-
Alexander Alekhin authored
-
arnaudbrejeon authored
-
Alexander Alekhin authored
-
Alexander Alekhin authored
-
Andrey Golubev authored
G-API: Use object instead of reference in extended lifetime case (#14946) * Use object instead of reference * Make the object const
-
Tomoaki Teshima authored
-
Alexander Alekhin authored
-
- 01 Jul, 2019 7 commits
-
-
armenpoghosov authored
Merge pull request #14828 from armenpoghosov:parmen_RANSACPointSetRegistrator_getSubset_disaster_cleanup Parmen ransac point set registrator get subset disaster cleanup (#14828)
-
Alexander Alekhin authored
-
Alexander Alekhin authored
-
Vitaly Tuzov authored
* Avoid using v_signmask universal intrinsic and mark it as deprecated * Renamed v_find_negative to v_scan_forward
-
Alexander Alekhin authored
-
Alexander Alekhin authored
-
Alexander Alekhin authored
-
- 30 Jun, 2019 2 commits
-
-
StefanBruens authored
Crosscorr cleanup (#14936) * Simplify code for convolution destination type/size For the 2d filter code, destination size equals source size, and the crossCorr function even (re-)creates the output matrix with the given size. The number of channels also have to match. The destination type() is the one used to create the output matrix, so we can use its type() here. This is a preparatory patch. Signed-off-by:
Stefan Brüns <stefan.bruens@rwth-aachen.de> * Remove redundant destination size and type parameters from crossCorr All calling sites of crossCorr already use (..., mat, mat.size(), mat.type(), ...), so the parameters are redundant. Signed-off-by:
Stefan Brüns <stefan.bruens@rwth-aachen.de>
-
Alexander Alekhin authored
-
- 29 Jun, 2019 7 commits
-
-
Stefan Brüns authored
Due to the explicitly declared copy constructor Vec<T, n>::Vec(Vec <T,n>&) GCC 9 warns if there is no assignment operator, as having one typically requires the other (rule-of-three, constructor/desctructor/assginment). As the values are just a plain array the default assignment operator does the right thing. Tell the compiler explicitly to default it. Signed-off-by:
Stefan Brüns <stefan.bruens@rwth-aachen.de>
-
Alexander Alekhin authored
-
Alexander Alekhin authored
-
Alexander Alekhin authored
-
Dmitry Kurtaev authored
-
Alexander Alekhin authored
- make it configurable via OPENCV_TBB_RELEASE + OPENCV_TBB_RELEASE_MD5 - remove legacy support
-
Alexander Alekhin authored
- builds with CMAKE_SKIP_RPATH=ON should properly find project .so files from <build>/lib directory - $ORIGIN doesn't work properly in cross-compilation mode
-
- 28 Jun, 2019 11 commits
-
-
Alexander Alekhin authored
-
Alexander Alekhin authored
-
Alexander Alekhin authored
-
Andrey Golubev authored
G-API: Align operators tests to new test model (#14922) * Align operators tests to new test model * Change init function in NotOperatorTest
-
Anton Potapov authored
- explanatory test cases for common usage models
-
atalaman authored
G-API: Parameterized render tests (#14892) * Init commit * Add mat size as test parameter * Add test for text render * Add test for rect render * Add tests for line and circle * Remove old render tests * Init output mats * Remove methods input arguments * Add comment about data loss in BGR2NV12 conversion * Add edge test cases * Replace default color for out mats black -> white
-
Stefan Brüns authored
In case OPENCV_INCLUDE_INSTALL_PATH is absolute (i.e. starts with a "/"), the path ends up with a double "/". While this is mostly equivalent to a single slash, it may have a nasty side effect when: - OpenCV_INSTALL_PATH is empty - OPENCV_INCLUDE_INSTALL_PATH is "/usr/include" - the calling build script uses "-isystem" to specify the path to the headers of dependencies (to avoid warnings) Specifying "-isystem /usr/include" breaks the path ordering, and GCC can no longer find its "stdlib.h", thus CMake filters such statements. Unfortunately it fails to do so when using "//usr/include".
-
Dmitry Kurtaev authored
-
Stefan Brüns authored
It is perfectly possible to compile ARM binaries on Aarch64. Generalize the code to no only swap X86 with X86_64, but also ARM and AARCH64. Signed-off-by:
Stefan Brüns <stefan.bruens@rwth-aachen.de>
-
Andrey Golubev authored
-
Andrey Golubev authored
G-API: Introduce new approach to write accuracy tests (#14757) * G-API: Introduce new common accuracy test fixture * Enable Range<> to Seq<> implicit conversion * Fix shadowing parameters * Update license headers * Rename ALIGNED_TYPE to SAME_TYPE * Move MkRange to tests * Fix TODO(agolubev) in test instantiations * Squash simple fixture declarations in one line * Remove unused line * Fix Windows issues with macro expansion * Choose between 1 or 2 matrix initialization * Redesign common class behavior Use "views" for GetParam() provided by GTest base class instead of doing segregation (with copy!) of common and specific parameters: request common or specific parameter directly by index from GetParam()-returned parameters * Refine user-level API and usage of new test model * Fix -fpermissive errors * Remove unnecessary init calls * Replace GCompileArgs member variable with func ptr * Rename initMatsRandN to make its behavior explicit Rename initMatsRandN to initMatrixRandN to eliminate confusion: initMatsRandN only initialized first matrix (similarly to initMatrixRandU) * Fix common of initNothing * Update copyright dates in missed files * Add check for specific parameters * Fix coment stlye
-