Also bring perf_imgproc CornerMinEigenVal accuracy requirements in line with
the test_imgproc accuracy requirements on that test and fix indentation on
the latter.
Partially addresses issue #9821
* goodFeaturesToTrack returns also corner value
(cherry picked from commit 4a8f06755c)
* Added response to GFTT Detector keypoints
(cherry picked from commit b88fb40c6e)
* Moved corner values to another optional variable to preserve backward compatibility
(cherry picked from commit 6137383d32)
* Removed corners valus from perf tests and better unit tests for corners values
(cherry picked from commit f3d0ef21a7)
* Fixed detector gftt call
(cherry picked from commit be2975553b)
* Restored test_cornerEigenValsVecs
(cherry picked from commit ea3e11811f)
* scaling fixed;
mineigen calculation rolled back;
gftt function overload added (with quality parameter);
perf tests were added for the new api function;
external bindings were added for the function (with different alias);
fixed issues with composition of the output array of the new function (e.g. as requested in comments) ;
added sanity checks in the perf tests;
removed C API changes.
* minor change to GFTTDetector::detect
* substitute ts->printf with EXPECT_LE
* avoid re-allocations
Co-authored-by: Anas <anas.el.amraoui@live.com>
Co-authored-by: amir.tulegenov <amir.tulegenov@xperience.ai>
Bit-exact Nearest Neighbor Resizing
* bit exact resizeNN
* change the value of method enum
* add bitexact-nn to ResizeExactTest
* test to compare with non-exact version
* add perf for bit-exact resizenn
* use cvFloor-equivalent
* 1/3 scaling is not stable for floating calculation
* stricter test
* bugfix: broken data in case of 6 or 12bytes elements
* bugfix: broken data in default pix_size
* stricter threshold
* use raw() for floor
* use double instead of int
* follow code reviews
* fewer cases in perf test
* center pixel convention
* resize: HResizeLinear reduce duplicate work
There appears to be a 2x unroll of the HResizeLinear against k,
however the k value is only incremented by 1 during the unroll. This
results in k - 1 duplicate passes when k > 1.
Likewise, the final pass may not respect the work done by the vector
loop. Start it with the offset returned by the vector op if
implemented. Note, no vector ops are implemented today.
The performance is most noticable on a linear downscale. A set of
performance tests are added to characterize this. The performance
improvement is 10-50% depending on the scaling.
* imgproc: vectorize HResizeLinear
Performance is mostly gated by the gather operations
for x inputs.
Likewise, provide a 2x unroll against k, this reduces the
number of alpha gathers by 1/2 for larger k.
While not a 4x improvement, it still performs substantially
better under P9 for a 1.4x improvement. P8 baseline is
1.05-1.10x due to reduced VSX instruction set.
For float types, this results in a more modest
1.2x improvement.
* Update U8 processing for non-bitexact linear resize
* core: hal: vsx: improve v_load_expand_q
With a little help, we can do this quickly without gprs on
all VSX enabled targets.
* resize: Fix cn == 3 step per feedback
Per feedback, ensure we don't overrun. This was caught via the
failure observed in Test_TensorFlow.inception_accuracy.
* Adding all possible data type interactions to the perf tests since some
use SIMD acceleration and others do not.
* Disabling full tests by default.
* Giving proper names, removing magic numbers and sanity checks of new
performance tests for the integral function.
* Giving proper names, making array static.
Due to size limit of shared memory, histogram is built on
the global memory for CV_16UC1 case.
The amount of memory needed for building histogram is:
65536 * 4byte = 256KB
and shared memory limit is 48KB typically.
Added test cases for CV_16UC1 and various clip limits.
Added perf tests for CV_16UC1 on both CPU and CUDA code.
There was also a bug in CV_8UC1 case when redistributing
"residual" clipped pixels. Adding the test case where clip
limit is 5.0 exposes this bug.
* added performance test for compareHist
* compareHist reworked to use wide universal intrinsics
* Disabled vectorization for CV_COMP_CORREL and CV_COMP_BHATTACHARYYA if f64 is unsupported
- improve cpu dispatching calls to allow more SIMD extentions
(SSE4.1, AVX2, VSX)
- wide universal intrinsics
- replace dummy v_expand with v_expand_low
- replace v_expand + v_mul_wrap with v_mul_expand for product accumulate operations
- use FMA for accumulate operations
- add mask and more types to accumulate's performance tests
* Add HPX backend for OpenCV implementation
Adds hpx backend for cv::parallel_for_() calls respecting the nstripes chunking parameter. C++ code for the backend is added to modules/core/parallel.cpp. Also, the necessary changes to cmake files are introduced.
Backend can operate in 2 versions (selectable by cmake build option WITH_HPX_STARTSTOP): hpx (runtime always on) and hpx_startstop (start and stop the backend for each cv::parallel_for_() call)
* WIP: Conditionally include hpx_main.hpp to tests in core module
Header hpx_main.hpp is included to both core/perf/perf_main.cpp and core/test/test_main.cpp.
The changes to cmake files for linking hpx library to above mentioned test executalbles are proposed but have issues.
* Add coditional iclusion of hpx_main.hpp to cpp cpu modules
* Remove start/stop version of hpx backend
* Added accumulator value to the output of HoughLines and HoughCircles
* imgproc: refactor Hough patch
- eliminate code duplication
- fix type handling, fix OpenCL code
- fix test data generation
- re-generated test data in debug mode via plain CPU code path
* Rewrite polar transformations
- A new wrapPolar function encapsulate both linear and semi-log remap
- Destination size is a parameter or calculated automatically to keep objects size between remapping
- linearPolar and logPolar has been deprecated
* Fix build warning and error in accuracy test
* Fix function name to warpPolar
* Explicitly specify the mapping mode, so we retain all the parameters as non-optional.
Introduces WarpPolarMode enum to specify the mapping mode in flags
* resolves performance warning on windows build
* removed duplicated logPolar and linearPolar implementations
* use universal intrinsic instead of raw intrinsic
* add 2 channels de-interleave on x86 platform
* add v_int32x4 version of v_muladd
* add accumulate version of v_dotprod based on the commit from seiko2plus on bf1852d
* remove some verify check in performance test
* avoid the out of boundary access and keep the performance
