]>
rtime.felk.cvut.cz Git - hercules2020/kcf.git/log
Shanigen [Thu, 24 May 2018 10:02:52 +0000 (12:02 +0200)]
Update README.md
Shanigen [Wed, 23 May 2018 12:18:53 +0000 (14:18 +0200)]
Added non-template version of the ComplexMat class to test out the HERCULES compiler.
Shanigen [Tue, 15 May 2018 17:22:37 +0000 (19:22 +0200)]
Corrected small mistake in scaling.
Shanigen [Tue, 15 May 2018 12:37:01 +0000 (14:37 +0200)]
Modified custom size window. You can now select both dimensions.
Shanigen [Mon, 14 May 2018 08:21:20 +0000 (10:21 +0200)]
Merge branch 'master' of https://github.com/Shanigen/kcf
Shanigen [Mon, 14 May 2018 08:20:26 +0000 (10:20 +0200)]
Corrected flags used during initialization of Fftw plans.
Shanigen [Sun, 13 May 2018 09:07:30 +0000 (11:07 +0200)]
Added source
Shanigen [Sat, 12 May 2018 13:05:32 +0000 (15:05 +0200)]
Modified parallel version of Fftw use POSIX threads only when the sequential version is used.
Shanigen [Wed, 9 May 2018 15:37:02 +0000 (17:37 +0200)]
Added sanity check to prevent combining OpenCV version with big batch mode, which it does not support.
Shanigen [Wed, 9 May 2018 15:16:35 +0000 (17:16 +0200)]
Addde warning, when using linear kernel with cufft version.
Shanigen [Wed, 9 May 2018 15:14:50 +0000 (17:14 +0200)]
Changed destructor of cuFFT to only destroy plans and delete data used in big batch mode, when big batch mode is on.
Shanigen [Wed, 9 May 2018 15:03:49 +0000 (17:03 +0200)]
Git corrected some issue, with destruction of fftw plans.
Shanigen [Wed, 9 May 2018 15:01:20 +0000 (17:01 +0200)]
Added support for big bach mode for linear kernal.
Shanigen [Wed, 9 May 2018 13:15:04 +0000 (15:15 +0200)]
Added sanity check for cuFFT version, which prevents unsupported combinations.
Shanigen [Wed, 9 May 2018 11:43:07 +0000 (13:43 +0200)]
Delete kcf.o
Shanigen [Wed, 9 May 2018 11:42:48 +0000 (13:42 +0200)]
Delete kcf.h.gch
Shanigen [Mon, 7 May 2018 03:44:14 +0000 (05:44 +0200)]
Merge branch 'master' of https://github.com/Shanigen/kcf
Shanigen [Mon, 7 May 2018 03:42:03 +0000 (05:42 +0200)]
Added custom scaling size option. The final window dimension size has to be divasible by cell size.
Shanigen [Fri, 4 May 2018 08:59:10 +0000 (10:59 +0200)]
Update README.md
Shanigen [Mon, 23 Apr 2018 14:07:41 +0000 (16:07 +0200)]
Changed scaling of window to be always 128x128 resulting in 32x32 matrice for feature descriptors.
Shanigen [Mon, 23 Apr 2018 12:59:36 +0000 (14:59 +0200)]
Gaussian correlation for CUFFT version is now on GPU. Also corrected << operator for CUDA ComplexMat
Shanigen [Thu, 19 Apr 2018 17:20:02 +0000 (19:20 +0200)]
Added forward_raw function to ffts in preparation to full cuda support.
Shanigen [Thu, 19 Apr 2018 16:30:03 +0000 (18:30 +0200)]
Corrected small mistake with inverse raw in cufft.
Shanigen [Thu, 19 Apr 2018 16:04:36 +0000 (18:04 +0200)]
Addition of inverse_raw in ftt for preparition of full cuda support. Moved cuda error check functions to cuda folder.
Shanigen [Thu, 19 Apr 2018 15:39:05 +0000 (17:39 +0200)]
Some minor corrections to variable in kcf.
Shanigen [Thu, 19 Apr 2018 15:18:47 +0000 (17:18 +0200)]
Added check in complexmat to prevent double freeing data on same pointer. Also added global variables with
dimensions for future use in cuda implementation.
Shanigen [Thu, 19 Apr 2018 11:29:08 +0000 (13:29 +0200)]
Added error cheking for CUDA API calls and CUDA kernels.
Shanigen [Thu, 19 Apr 2018 09:33:55 +0000 (11:33 +0200)]
Removed unused cuda streams in cufft.
Michal Sojka [Mon, 16 Apr 2018 13:48:48 +0000 (15:48 +0200)]
Merge branch 'master' of github.com:Shanigen/kcf
Michal Sojka [Mon, 16 Apr 2018 13:48:44 +0000 (15:48 +0200)]
Fix indentation and whitespace
Shanigen [Mon, 16 Apr 2018 11:56:09 +0000 (13:56 +0200)]
Added safety check for image dimensions in cuFFT version.
Shanigen [Tue, 10 Apr 2018 13:05:15 +0000 (15:05 +0200)]
Merge branch 'master' of https://github.com/Shanigen/kcf
Shanigen [Tue, 10 Apr 2018 13:00:23 +0000 (15:00 +0200)]
Removed zero copy for complexmat now it is using normal device memory. Added zero copy memory to cufft and minimized memory copy from device to host and host to device.
Michal Sojka [Mon, 9 Apr 2018 11:14:25 +0000 (13:14 +0200)]
Do not delete OpenCV_DIR example from toplevel Makefile
Shanigen [Fri, 6 Apr 2018 09:54:38 +0000 (11:54 +0200)]
Modified Makefile to build all the current working options.
Shanigen [Wed, 4 Apr 2018 18:12:44 +0000 (20:12 +0200)]
Completed implementation of cufft and complexmat cuda implementation with custom kernels. Also added big batch variable to ffts.
Shanigen [Thu, 29 Mar 2018 11:08:36 +0000 (13:08 +0200)]
Sqr norm in complexmat now returns array.
Shanigen [Wed, 28 Mar 2018 14:54:11 +0000 (16:54 +0200)]
Deleted obsolete code.
Shanigen [Wed, 28 Mar 2018 13:03:05 +0000 (15:03 +0200)]
Removed custom allocator in favor of thrust.
Shanigen [Wed, 28 Mar 2018 08:37:56 +0000 (10:37 +0200)]
Added custom allocator for vectors in ComplexMat when using cuFFT.
Shanigen [Wed, 28 Mar 2018 08:14:27 +0000 (10:14 +0200)]
Corrected init in OpenCV FFT version
Shanigen [Mon, 26 Mar 2018 15:41:42 +0000 (17:41 +0200)]
Added compatibility for Opencv 3 in cuFFT.
Shanigen [Mon, 26 Mar 2018 14:10:26 +0000 (16:10 +0200)]
Added cmake support for cufft, also copied fftw functions to cufft and changed the to support cufft(in progress).
Shanigen [Mon, 26 Mar 2018 11:20:39 +0000 (13:20 +0200)]
cuFFT use preparation correction
Shanigen [Mon, 26 Mar 2018 11:17:11 +0000 (13:17 +0200)]
Prepare using cuFFT
Shanigen [Mon, 26 Mar 2018 08:34:34 +0000 (10:34 +0200)]
Corrected small mistake with OpenMP support for big batch mode.
Shanigen [Mon, 26 Mar 2018 08:35:50 +0000 (10:35 +0200)]
Added support for OMP in big batch mode.
Shanigen [Sun, 25 Mar 2018 16:51:50 +0000 (18:51 +0200)]
Big batch mode is now working correctly.
Shanigen [Sun, 25 Mar 2018 15:27:32 +0000 (17:27 +0200)]
Big batch update 2 (not finished)
Problem with scaling.
Shanigen [Wed, 21 Mar 2018 18:52:09 +0000 (19:52 +0100)]
Big batch update (not finished)
Shanigen [Tue, 20 Mar 2018 15:26:34 +0000 (16:26 +0100)]
Added number of scales to FFT init.
Shanigen [Tue, 20 Mar 2018 15:18:12 +0000 (16:18 +0100)]
Added number of features to FFT init.
Shanigen [Tue, 20 Mar 2018 09:29:15 +0000 (10:29 +0100)]
FFTW plans use FFTW_MEASURE now
Shanigen [Tue, 20 Mar 2018 08:57:58 +0000 (09:57 +0100)]
FFTW cleanup version 1. done.
Michal Sojka [Mon, 19 Mar 2018 14:09:10 +0000 (15:09 +0100)]
fftw cleanup (not finished)
Shanigen [Mon, 19 Mar 2018 10:19:38 +0000 (11:19 +0100)]
Removed debug couts.
Shanigen [Mon, 19 Mar 2018 10:17:13 +0000 (11:17 +0100)]
ComplexMat is now output of forward_window fft.
Shanigen [Sun, 18 Mar 2018 12:39:55 +0000 (13:39 +0100)]
Changed p_data in ComplexMat to mutable so it can be modified using the get_p_data const function.
Shanigen [Fri, 16 Mar 2018 13:41:05 +0000 (14:41 +0100)]
Added Big Batch mode for forward_window, which enables to transforming all features from all scales together.
Shanigen [Thu, 15 Mar 2018 16:16:27 +0000 (17:16 +0100)]
Added simpler cufftw fix also to the 1 channel iFFT.
Shanigen [Thu, 15 Mar 2018 09:38:09 +0000 (10:38 +0100)]
Merge branch 'master' of github.com:Shanigen/kcf
Shanigen [Thu, 15 Mar 2018 09:37:15 +0000 (10:37 +0100)]
Changed ComplexMat p_data from vector of vectors of complex numbers to vector of complex numbers.
Michal Sojka [Wed, 14 Mar 2018 22:07:50 +0000 (23:07 +0100)]
Merge branch 'master' of github.com:Shanigen/kcf
Shanigen [Tue, 13 Mar 2018 10:47:06 +0000 (11:47 +0100)]
Corrected small mistake.
Shanigen [Tue, 13 Mar 2018 10:43:18 +0000 (11:43 +0100)]
Added New-array Execute function support for FFTW.
Shanigen [Tue, 13 Mar 2018 08:24:29 +0000 (09:24 +0100)]
FFTW OpenMP support restored.
Michal Sojka [Mon, 12 Mar 2018 23:43:35 +0000 (00:43 +0100)]
Simplify cufftw fix
There is no need to special-case cuda and non-cuda versions.
Shanigen [Mon, 12 Mar 2018 22:35:05 +0000 (23:35 +0100)]
Added cmake modification for fftw async.
Shanigen [Mon, 12 Mar 2018 13:45:22 +0000 (14:45 +0100)]
FFTW async support with multi threaded plans restored.
Shanigen [Mon, 12 Mar 2018 13:08:12 +0000 (14:08 +0100)]
Corrected cuFFTW inverse FFT.
Michal Sojka [Sun, 11 Mar 2018 22:55:19 +0000 (23:55 +0100)]
Finish conversion
There seems to be some bug in this commit. On TX2, tracking
performance is bad. "Native FFTW" seems to be OK.
Michal Sojka [Sun, 11 Mar 2018 20:22:58 +0000 (21:22 +0100)]
Start converting fftw version to "plan_many"
Michal Sojka [Sun, 11 Mar 2018 08:09:53 +0000 (09:09 +0100)]
Add DEBUG_PRINTF macro
Michal Sojka [Sun, 11 Mar 2018 13:42:20 +0000 (14:42 +0100)]
Do not expect that size of fourier transform input and output is the same
FFTW computes only a half of the full result, because is other half is
complex conjugate of the first half.
Michal Sojka [Sun, 11 Mar 2018 12:45:02 +0000 (13:45 +0100)]
Add fftw-based FFT implementation
Michal Sojka [Sun, 11 Mar 2018 12:15:09 +0000 (13:15 +0100)]
Introduce FFT abstraction and move implementation to separate files
Also move cmake FFT logic from root directory to KCF library in src.
Michal Sojka [Sun, 11 Mar 2018 08:03:18 +0000 (09:03 +0100)]
Make downscaling configurable
This is useful for debugging - smaller images are easier to print on
terminal.
Michal Sojka [Sun, 11 Mar 2018 07:59:31 +0000 (08:59 +0100)]
cmake: Make it easy to override -O3 with CMAKE_BUILD_TYPE
Michal Sojka [Fri, 9 Mar 2018 23:53:31 +0000 (00:53 +0100)]
Remove profiling code
In its current form, it complicates understanding of the rest of the
code.
Michal Sojka [Wed, 7 Mar 2018 18:25:59 +0000 (19:25 +0100)]
cmake: Use Ninja generator rather than Makefile
Ninja is able to better paralelize the build and thus saves time when
experimenting with builds.
Michal Sojka [Wed, 7 Mar 2018 18:08:31 +0000 (19:08 +0100)]
cmake: Move big part of cmake logic to src/
The reason is that it is the KCF library (in src/), which needs this logic,
not the main program using the library.
Shanigen [Tue, 6 Mar 2018 16:04:45 +0000 (17:04 +0100)]
Merge branch 'master' of https://github.com/Shanigen/kcf
Shanigen [Tue, 6 Mar 2018 15:47:39 +0000 (16:47 +0100)]
Corrected "corrected" fftw implementation, also made it in-place.
Michal Sojka [Tue, 6 Mar 2018 11:24:15 +0000 (12:24 +0100)]
Remove wrong call to cv::dft in fftw version
Michal Sojka [Tue, 6 Mar 2018 10:09:02 +0000 (11:09 +0100)]
Make debug run-time option rather than compile time
Shanigen [Tue, 6 Mar 2018 10:29:57 +0000 (11:29 +0100)]
Corrected scaling issue in inverse fftw.
Michal Sojka [Tue, 6 Mar 2018 07:46:02 +0000 (08:46 +0100)]
cmake: Find CUDA before OpenCV
This way, CUDA is searched only once. OpenCV seems to not search it again, if
it is already found.
Michal Sojka [Tue, 6 Mar 2018 07:41:01 +0000 (08:41 +0100)]
Makefile: Add shortcut targets
Now, one can run just 'make cufftw' to compile the particular variant.
Michal Sojka [Tue, 6 Mar 2018 07:40:39 +0000 (08:40 +0100)]
cmake: Find CUDA before going into src
This fixes few compilation issues on TX2
Shanigen [Mon, 5 Mar 2018 17:15:03 +0000 (18:15 +0100)]
Added support for cuFFTW and corrected wrong usage of OpenMP in FFTW.
Michal Sojka [Mon, 5 Mar 2018 16:09:34 +0000 (17:09 +0100)]
Fix linking of cufftw library
Michal Sojka [Mon, 5 Mar 2018 15:43:53 +0000 (16:43 +0100)]
Pass proper compiler options for cufftw compilation
Michal Sojka [Mon, 5 Mar 2018 15:18:39 +0000 (16:18 +0100)]
Do not use external project for libpfm
It's simpler to run 'apt install libpfm4-dev' and everything works.
Shanigen [Sun, 4 Mar 2018 18:26:23 +0000 (19:26 +0100)]
Added profiling option to the project using perf_event_open and libpfm4. Also corrected some mistakes in README and corrected possible issue if FFTW async version.
Shanigen [Fri, 2 Mar 2018 22:37:26 +0000 (23:37 +0100)]
Added async support to fftw and deleted some obsolete code in fhog.hpp.
Michal Sojka [Fri, 2 Mar 2018 14:06:58 +0000 (15:06 +0100)]
Prepare for using cuFFTW
This does not work yet, as we currently use cuFFTW unsupported functions
such as fftw_alloc_real.
Michal Sojka [Fri, 2 Mar 2018 14:05:23 +0000 (15:05 +0100)]
Fix placement of omp critical
Michal Sojka [Fri, 2 Mar 2018 13:49:27 +0000 (14:49 +0100)]
Better way to not warn about omp pragmas when compiling without omp
Michal Sojka [Fri, 2 Mar 2018 13:02:23 +0000 (14:02 +0100)]
Add --output parameter
Michal Sojka [Fri, 2 Mar 2018 12:55:03 +0000 (13:55 +0100)]
Automatically use groundtruth.txt when it exists (with zero or one arguments)