mirror/opencv
1
0
Fork 0
mirror of https://github.com/opencv/opencv.git synced 2024-11-28 21:20:18 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

revise accuracy and perf tests

Browse Source
This commit is contained in:
yao 2013-06-19 13:03:35 +08:00
parent 843094a07f
commit 2c198f6cd6
14 changed files with 392 additions and 614 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
modules/ocl/perf/main.cpp
View File

@ -52,6 +52,8 @@ int main(int argc, const char *argv[])
cerr << "no device found\n";
return -1;
}
// set this to overwrite binary cache every time the test starts
ocl::setBinaryDiskCache(ocl::CACHE_UPDATE);
int devidx = 0;

85
modules/ocl/test/test_columnsum.cpp → modules/ocl/perf/perf_calib3d.cpp
View File

@ -15,8 +15,8 @@
// Third party copyrights are property of their respective owners.
//
// @Authors
// Chunpeng Zhang chunpeng@multicorewareinc.com
//
// Fangfang Bai, fangfang@multicorewareinc.com
// Jin Ma, jin@multicorewareinc.com
//
// Redistribution and use in source and binary forms, with or without modification,
// are permitted provided that the following conditions are met:
@ -31,7 +31,7 @@
// * The name of the copyright holders may not be used to endorse or promote products
// derived from this software without specific prior written permission.
//
// This software is provided by the copyright holders and contributors "as is" and
// This software is provided by the copyright holders and contributors as is and
// any express or implied warranties, including, but not limited to, the implied
// warranties of merchantability and fitness for a particular purpose are disclaimed.
// In no event shall the Intel Corporation or contributors be liable for any direct,
@ -45,50 +45,57 @@
//M*/
#include "precomp.hpp"
#include <iomanip>
#ifdef HAVE_OPENCL
PARAM_TEST_CASE(ColumnSum, cv::Size)
///////////// StereoMatchBM ////////////////////////
PERFTEST(StereoMatchBM)
{
cv::Size size;
cv::Mat src;
Mat left_image = imread(abspath("aloeL.jpg"), cv::IMREAD_GRAYSCALE);
Mat right_image = imread(abspath("aloeR.jpg"), cv::IMREAD_GRAYSCALE);
Mat disp,dst;
ocl::oclMat d_left, d_right,d_disp;
int n_disp= 128;
int winSize =19;
virtual void SetUp()
{
size = GET_PARAM(0);
}
};
SUBTEST << left_image.cols << 'x' << left_image.rows << "; aloeL.jpg ;"<< right_image.cols << 'x' << right_image.rows << "; aloeR.jpg ";
TEST_P(ColumnSum, Accuracy)
{
cv::Mat src = randomMat(size, CV_32FC1);
cv::ocl::oclMat d_dst;
cv::ocl::oclMat d_src(src);
StereoBM bm(0, n_disp, winSize);
bm(left_image, right_image, dst);
cv::ocl::columnSum(d_src, d_dst);
CPU_ON;
bm(left_image, right_image, dst);
CPU_OFF;
cv::Mat dst(d_dst);
d_left.upload(left_image);
d_right.upload(right_image);
for (int j = 0; j < src.cols; ++j)
{
float gold = src.at<float>(0, j);
float res = dst.at<float>(0, j);
ASSERT_NEAR(res, gold, 1e-5);
}
ocl::StereoBM_OCL d_bm(0, n_disp, winSize);
for (int i = 1; i < src.rows; ++i)
{
for (int j = 0; j < src.cols; ++j)
{
float gold = src.at<float>(i, j) += src.at<float>(i - 1, j);
float res = dst.at<float>(i, j);
ASSERT_NEAR(res, gold, 1e-5);
}
}
WARMUP_ON;
d_bm(d_left, d_right, d_disp);
WARMUP_OFF;
cv::Mat ocl_mat;
d_disp.download(ocl_mat);
ocl_mat.convertTo(ocl_mat, dst.type());
GPU_ON;
d_bm(d_left, d_right, d_disp);
GPU_OFF;
GPU_FULL_ON;
d_left.upload(left_image);
d_right.upload(right_image);
d_bm(d_left, d_right, d_disp);
d_disp.download(disp);
GPU_FULL_OFF;
TestSystem::instance().setAccurate(-1, 0.);
}
INSTANTIATE_TEST_CASE_P(OCL_ImgProc, ColumnSum, DIFFERENT_SIZES);
#endif

16
modules/ocl/perf/perf_filters.cpp
View File

@ -284,6 +284,7 @@ PERFTEST(GaussianBlur)
Mat src, dst, ocl_dst;
int all_type[] = {CV_8UC1, CV_8UC4, CV_32FC1, CV_32FC4};
std::string type_name[] = {"CV_8UC1", "CV_8UC4", "CV_32FC1", "CV_32FC4"};
const int ksize = 7;
for (int size = Min_Size; size <= Max_Size; size *= Multiple)
{
@ -291,29 +292,28 @@ PERFTEST(GaussianBlur)
{
SUBTEST << size << 'x' << size << "; " << type_name[j] ;
gen(src, size, size, all_type[j], 5, 16);
gen(src, size, size, all_type[j], 0, 256);
GaussianBlur(src, dst, Size(9, 9), 0);
GaussianBlur(src, dst, Size(ksize, ksize), 0);
CPU_ON;
GaussianBlur(src, dst, Size(9, 9), 0);
GaussianBlur(src, dst, Size(ksize, ksize), 0);
CPU_OFF;
ocl::oclMat d_src(src);
ocl::oclMat d_dst(src.size(), src.type());
ocl::oclMat d_buf;
ocl::oclMat d_dst;
WARMUP_ON;
ocl::GaussianBlur(d_src, d_dst, Size(9, 9), 0);
ocl::GaussianBlur(d_src, d_dst, Size(ksize, ksize), 0);
WARMUP_OFF;
GPU_ON;
ocl::GaussianBlur(d_src, d_dst, Size(9, 9), 0);
ocl::GaussianBlur(d_src, d_dst, Size(ksize, ksize), 0);
GPU_OFF;
GPU_FULL_ON;
d_src.upload(src);
ocl::GaussianBlur(d_src, d_dst, Size(9, 9), 0);
ocl::GaussianBlur(d_src, d_dst, Size(ksize, ksize), 0);
d_dst.download(ocl_dst);
GPU_FULL_OFF;

76
modules/ocl/perf/perf_hog.cpp
View File

@ -46,11 +46,6 @@
#include "precomp.hpp"
///////////// HOG////////////////////////
bool match_rect(cv::Rect r1, cv::Rect r2, int threshold)
{
return ((abs(r1.x - r2.x) < threshold) && (abs(r1.y - r2.y) < threshold) &&
(abs(r1.width - r2.width) < threshold) && (abs(r1.height - r2.height) < threshold));
}
PERFTEST(HOG)
{
@ -61,13 +56,12 @@ PERFTEST(HOG)
throw runtime_error("can't open road.png");
}
cv::HOGDescriptor hog;
hog.setSVMDetector(hog.getDefaultPeopleDetector());
std::vector<cv::Rect> found_locations;
std::vector<cv::Rect> d_found_locations;
SUBTEST << 768 << 'x' << 576 << "; road.png";
SUBTEST << src.cols << 'x' << src.rows << "; road.png";
hog.detectMultiScale(src, found_locations);
@ -84,70 +78,10 @@ PERFTEST(HOG)
ocl_hog.detectMultiScale(d_src, d_found_locations);
WARMUP_OFF;
// Ground-truth rectangular people window
cv::Rect win1_64x128(231, 190, 72, 144);
cv::Rect win2_64x128(621, 156, 97, 194);
cv::Rect win1_48x96(238, 198, 63, 126);
cv::Rect win2_48x96(619, 161, 92, 185);
cv::Rect win3_48x96(488, 136, 56, 112);
// Compare whether ground-truth windows are detected and compare the number of windows detected.
std::vector<int> d_comp(4);
std::vector<int> comp(4);
for(int i = 0; i < (int)d_comp.size(); i++)
{
d_comp[i] = 0;
comp[i] = 0;
}
int threshold = 10;
int val = 32;
d_comp[0] = (int)d_found_locations.size();
comp[0] = (int)found_locations.size();
cv::Size winSize = hog.winSize;
if (winSize == cv::Size(48, 96))
{
for(int i = 0; i < (int)d_found_locations.size(); i++)
{
if (match_rect(d_found_locations[i], win1_48x96, threshold))
d_comp[1] = val;
if (match_rect(d_found_locations[i], win2_48x96, threshold))
d_comp[2] = val;
if (match_rect(d_found_locations[i], win3_48x96, threshold))
d_comp[3] = val;
}
for(int i = 0; i < (int)found_locations.size(); i++)
{
if (match_rect(found_locations[i], win1_48x96, threshold))
comp[1] = val;
if (match_rect(found_locations[i], win2_48x96, threshold))
comp[2] = val;
if (match_rect(found_locations[i], win3_48x96, threshold))
comp[3] = val;
}
}
else if (winSize == cv::Size(64, 128))
{
for(int i = 0; i < (int)d_found_locations.size(); i++)
{
if (match_rect(d_found_locations[i], win1_64x128, threshold))
d_comp[1] = val;
if (match_rect(d_found_locations[i], win2_64x128, threshold))
d_comp[2] = val;
}
for(int i = 0; i < (int)found_locations.size(); i++)
{
if (match_rect(found_locations[i], win1_64x128, threshold))
comp[1] = val;
if (match_rect(found_locations[i], win2_64x128, threshold))
comp[2] = val;
}
}
cv::Mat gpu_rst(d_comp), cpu_rst(comp);
TestSystem::instance().ExpectedMatNear(gpu_rst, cpu_rst, 3);
if(d_found_locations.size() == found_locations.size())
TestSystem::instance().setAccurate(1, 0);
else
TestSystem::instance().setAccurate(0, abs((int)found_locations.size() - (int)d_found_locations.size()));
GPU_ON;
ocl_hog.detectMultiScale(d_src, found_locations);

46
modules/ocl/perf/perf_imgproc.cpp
View File

@ -743,12 +743,12 @@ PERFTEST(meanShiftFiltering)
WARMUP_OFF;
GPU_ON;
ocl::meanShiftFiltering(d_src, d_dst, sp, sr);
ocl::meanShiftFiltering(d_src, d_dst, sp, sr, crit);
GPU_OFF;
GPU_FULL_ON;
d_src.upload(src);
ocl::meanShiftFiltering(d_src, d_dst, sp, sr);
ocl::meanShiftFiltering(d_src, d_dst, sp, sr, crit);
d_dst.download(ocl_dst);
GPU_FULL_OFF;
@ -969,3 +969,45 @@ PERFTEST(CLAHE)
}
}
}
///////////// columnSum////////////////////////
PERFTEST(columnSum)
{
Mat src, dst, ocl_dst;
ocl::oclMat d_src, d_dst;
for (int size = Min_Size; size <= Max_Size; size *= Multiple)
{
SUBTEST << size << 'x' << size << "; CV_32FC1";
gen(src, size, size, CV_32FC1, 0, 256);
CPU_ON;
dst.create(src.size(), src.type());
for (int j = 0; j < src.cols; j++)
dst.at<float>(0, j) = src.at<float>(0, j);
for (int i = 1; i < src.rows; ++i)
for (int j = 0; j < src.cols; ++j)
dst.at<float>(i, j) = dst.at<float>(i - 1 , j) + src.at<float>(i , j);
CPU_OFF;
d_src.upload(src);
WARMUP_ON;
ocl::columnSum(d_src, d_dst);
WARMUP_OFF;
GPU_ON;
ocl::columnSum(d_src, d_dst);
GPU_OFF;
GPU_FULL_ON;
d_src.upload(src);
ocl::columnSum(d_src, d_dst);
d_dst.download(ocl_dst);
GPU_FULL_OFF;
TestSystem::instance().ExpectedMatNear(dst, ocl_dst, 5e-1);
}
}

60
modules/ocl/perf/perf_columnsum.cpp → modules/ocl/perf/perf_moments.cpp
View File

@ -44,45 +44,49 @@
//
//M*/
#include "precomp.hpp"
///////////// columnSum////////////////////////
PERFTEST(columnSum)
///////////// Moments ////////////////////////
PERFTEST(Moments)
{
Mat src, dst, ocl_dst;
ocl::oclMat d_src, d_dst;
Mat src;
bool binaryImage = 0;
int all_type[] = {CV_8UC1, CV_16SC1, CV_32FC1, CV_64FC1};
std::string type_name[] = {"CV_8UC1", "CV_16SC1", "CV_32FC1", "CV_64FC1"};
for (int size = Min_Size; size <= Max_Size; size *= Multiple)
{
SUBTEST << size << 'x' << size << "; CV_32FC1";
for (size_t j = 0; j < sizeof(all_type) / sizeof(int); j++)
{
SUBTEST << size << 'x' << size << "; " << type_name[j];
gen(src, size, size, CV_32FC1, 0, 256);
gen(src, size, size, all_type[j], 0, 256);
CPU_ON;
dst.create(src.size(), src.type());
for (int j = 0; j < src.cols; j++)
dst.at<float>(0, j) = src.at<float>(0, j);
cv::Moments CvMom = moments(src, binaryImage);
for (int i = 1; i < src.rows; ++i)
for (int j = 0; j < src.cols; ++j)
dst.at<float>(i, j) = dst.at<float>(i - 1 , j) + src.at<float>(i , j);
CPU_OFF;
CPU_ON;
moments(src, binaryImage);
CPU_OFF;
d_src.upload(src);
cv::Moments oclMom;
WARMUP_ON;
oclMom = ocl::ocl_moments(src, binaryImage);
WARMUP_OFF;
WARMUP_ON;
ocl::columnSum(d_src, d_dst);
WARMUP_OFF;
Mat gpu_dst, cpu_dst;
HuMoments(CvMom, cpu_dst);
HuMoments(oclMom, gpu_dst);
GPU_ON;
ocl::columnSum(d_src, d_dst);
GPU_OFF;
GPU_ON;
ocl::ocl_moments(src, binaryImage);
GPU_OFF;
GPU_FULL_ON;
d_src.upload(src);
ocl::columnSum(d_src, d_dst);
d_dst.download(ocl_dst);
GPU_FULL_OFF;
GPU_FULL_ON;
ocl::ocl_moments(src, binaryImage);
GPU_FULL_OFF;
TestSystem::instance().ExpectedMatNear(gpu_dst, cpu_dst, .5);
}
TestSystem::instance().ExpectedMatNear(dst, ocl_dst, 5e-1);
}
}

14
modules/ocl/perf/precomp.cpp
View File

@ -331,20 +331,6 @@ void TestSystem::printMetrics(int is_accurate, double cpu_time, double gpu_time,
cout << setiosflags(ios_base::left);
stringstream stream;
#if 0
if(is_accurate == 1)
stream << "Pass";
else if(is_accurate_ == 0)
stream << "Fail";
else if(is_accurate == -1)
stream << " ";
else
{
std::cout<<"is_accurate errer: "<<is_accurate<<"\n";
exit(-1);
}
#endif
std::stringstream &cur_subtest_description = getCurSubtestDescription();
#if GTEST_OS_WINDOWS&&!GTEST_OS_WINDOWS_MOBILE

180
modules/ocl/test/test_haar.cpp
View File

@ -1,180 +0,0 @@
/*M///////////////////////////////////////////////////////////////////////////////////////
//
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
//
// By downloading, copying, installing or using the software you agree to this license.
// If you do not agree to this license, do not download, install,
// copy or use the software.
//
//
// License Agreement
// For Open Source Computer Vision Library
//
// Copyright (C) 2010-2012, Institute Of Software Chinese Academy Of Science, all rights reserved.
// Copyright (C) 2010-2012, Advanced Micro Devices, Inc., all rights reserved.
// Copyright (C) 2010-2012, Multicoreware, Inc., all rights reserved.
// Third party copyrights are property of their respective owners.
//
// @Authors
// Jia Haipeng, jiahaipeng95@gmail.com
// Sen Liu, swjutls1987@126.com
//
// Redistribution and use in source and binary forms, with or without modification,
// are permitted provided that the following conditions are met:
//
// * Redistribution's of source code must retain the above copyright notice,
// this list of conditions and the following disclaimer.
//
// * Redistribution's in binary form must reproduce the above copyright notice,
// this list of conditions and the following disclaimer in the documentation
// and/or other oclMaterials provided with the distribution.
//
// * The name of the copyright holders may not be used to endorse or promote products
// derived from this software without specific prior written permission.
//
// This software is provided by the copyright holders and contributors "as is" and
// any express or implied warranties, including, but not limited to, the implied
// warranties of merchantability and fitness for a particular purpose are disclaimed.
// In no event shall the Intel Corporation or contributors be liable for any direct,
// indirect, incidental, special, exemplary, or consequential damages
// (including, but not limited to, procurement of substitute goods or services;
// loss of use, data, or profits; or business interruption) however caused
// and on any theory of liability, whether in contract, strict liability,
// or tort (including negligence or otherwise) arising in any way out of
// the use of this software, even if advised of the possibility of such damage.
//
//M*/
#include "opencv2/objdetect/objdetect.hpp"
#include "precomp.hpp"
#ifdef HAVE_OPENCL
using namespace cvtest;
using namespace testing;
using namespace std;
using namespace cv;
extern string workdir;
namespace
{
IMPLEMENT_PARAM_CLASS(CascadeName, std::string);
CascadeName cascade_frontalface_alt(std::string("haarcascade_frontalface_alt.xml"));
CascadeName cascade_frontalface_alt2(std::string("haarcascade_frontalface_alt2.xml"));
struct getRect
{
Rect operator ()(const CvAvgComp &e) const
{
return e.rect;
}
};
}
PARAM_TEST_CASE(Haar, double, int, CascadeName)
{
cv::ocl::OclCascadeClassifier cascade, nestedCascade;
cv::CascadeClassifier cpucascade, cpunestedCascade;
double scale;
int flags;
std::string cascadeName;
virtual void SetUp()
{
scale = GET_PARAM(0);
flags = GET_PARAM(1);
cascadeName = (workdir + "../../data/haarcascades/").append(GET_PARAM(2));
if( (!cascade.load( cascadeName )) || (!cpucascade.load(cascadeName)) )
{
cout << "ERROR: Could not load classifier cascade" << endl;
return;
}
}
};
////////////////////////////////faceDetect/////////////////////////////////////////////////
TEST_P(Haar, FaceDetect)
{
string imgName = workdir + "lena.jpg";
Mat img = imread( imgName, 1 );
if(img.empty())
{
std::cout << "Couldn't read " << imgName << std::endl;
return ;
}
vector<Rect> faces, oclfaces;
Mat gray, smallImg(cvRound (img.rows / scale), cvRound(img.cols / scale), CV_8UC1 );
MemStorage storage(cvCreateMemStorage(0));
cvtColor( img, gray, CV_BGR2GRAY );
resize( gray, smallImg, smallImg.size(), 0, 0, INTER_LINEAR );
equalizeHist( smallImg, smallImg );
cv::ocl::oclMat image;
CvSeq *_objects;
image.upload(smallImg);
_objects = cascade.oclHaarDetectObjects( image, storage, 1.1,
3, flags, Size(30, 30), Size(0, 0) );
vector<CvAvgComp> vecAvgComp;
Seq<CvAvgComp>(_objects).copyTo(vecAvgComp);
oclfaces.resize(vecAvgComp.size());
std::transform(vecAvgComp.begin(), vecAvgComp.end(), oclfaces.begin(), getRect());
cpucascade.detectMultiScale( smallImg, faces, 1.1, 3,
flags,
Size(30, 30), Size(0, 0) );
EXPECT_EQ(faces.size(), oclfaces.size());
}
TEST_P(Haar, FaceDetectUseBuf)
{
string imgName = workdir + "lena.jpg";
Mat img = imread( imgName, 1 );
if(img.empty())
{
std::cout << "Couldn't read " << imgName << std::endl;
return ;
}
vector<Rect> faces, oclfaces;
Mat gray, smallImg(cvRound (img.rows / scale), cvRound(img.cols / scale), CV_8UC1 );
cvtColor( img, gray, CV_BGR2GRAY );
resize( gray, smallImg, smallImg.size(), 0, 0, INTER_LINEAR );
equalizeHist( smallImg, smallImg );
cv::ocl::oclMat image;
image.upload(smallImg);
cv::ocl::OclCascadeClassifierBuf cascadebuf;
if( !cascadebuf.load( cascadeName ) )
{
cout << "ERROR: Could not load classifier cascade for FaceDetectUseBuf!" << endl;
return;
}
cascadebuf.detectMultiScale( image, oclfaces, 1.1, 3,
flags,
Size(30, 30), Size(0, 0) );
cpucascade.detectMultiScale( smallImg, faces, 1.1, 3,
flags,
Size(30, 30), Size(0, 0) );
EXPECT_EQ(faces.size(), oclfaces.size());
// intentionally run ocl facedetect again and check if it still works after the first run
cascadebuf.detectMultiScale( image, oclfaces, 1.1, 3,
flags,
Size(30, 30));
cascadebuf.release();
EXPECT_EQ(faces.size(), oclfaces.size());
}
INSTANTIATE_TEST_CASE_P(FaceDetect, Haar,
Combine(Values(1.0),
Values(CV_HAAR_SCALE_IMAGE, 0), Values(cascade_frontalface_alt, cascade_frontalface_alt2)));
#endif // HAVE_OPENCL

46
modules/ocl/test/test_imgproc.cpp
View File

@ -1573,6 +1573,47 @@ TEST_P(Convolve, Mat)
}
}
//////////////////////////////// ColumnSum //////////////////////////////////////
PARAM_TEST_CASE(ColumnSum, cv::Size)
{
cv::Size size;
cv::Mat src;
virtual void SetUp()
{
size = GET_PARAM(0);
}
};
TEST_P(ColumnSum, Accuracy)
{
cv::Mat src = randomMat(size, CV_32FC1);
cv::ocl::oclMat d_dst;
cv::ocl::oclMat d_src(src);
cv::ocl::columnSum(d_src, d_dst);
cv::Mat dst(d_dst);
for (int j = 0; j < src.cols; ++j)
{
float gold = src.at<float>(0, j);
float res = dst.at<float>(0, j);
ASSERT_NEAR(res, gold, 1e-5);
}
for (int i = 1; i < src.rows; ++i)
{
for (int j = 0; j < src.cols; ++j)
{
float gold = src.at<float>(i, j) += src.at<float>(i - 1, j);
float res = dst.at<float>(i, j);
ASSERT_NEAR(res, gold, 1e-5);
}
}
}
/////////////////////////////////////////////////////////////////////////////////////
INSTANTIATE_TEST_CASE_P(ImgprocTestBase, equalizeHist, Combine(
ONE_TYPE(CV_8UC1),
NULL_TYPE,
@ -1688,7 +1729,6 @@ INSTANTIATE_TEST_CASE_P(ImgProc, CLAHE, Combine(
Values(cv::Size(128, 128), cv::Size(113, 113), cv::Size(1300, 1300)),
Values(0.0, 40.0)));
//INSTANTIATE_TEST_CASE_P(ConvolveTestBase, Convolve, Combine(
// Values(CV_32FC1, CV_32FC1),
// Values(false))); // Values(false) is the reserved parameter
INSTANTIATE_TEST_CASE_P(OCL_ImgProc, ColumnSum, DIFFERENT_SIZES);
#endif // HAVE_OPENCL

229
modules/ocl/test/test_hog.cpp → modules/ocl/test/test_objdetect.cpp
View File

@ -15,7 +15,7 @@
// Third party copyrights are property of their respective owners.
//
// @Authors
// Wenju He, wenju@multicorewareinc.com
// Yao Wang, bitwangyaoyao@gmail.com
//
// Redistribution and use in source and binary forms, with or without modification,
// are permitted provided that the following conditions are met:
@ -45,51 +45,58 @@
#include "precomp.hpp"
#include "opencv2/core/core.hpp"
using namespace std;
#include "opencv2/objdetect/objdetect.hpp"
using namespace cv;
using namespace testing;
#ifdef HAVE_OPENCL
extern string workdir;
PARAM_TEST_CASE(HOG, cv::Size, int)
///////////////////// HOG /////////////////////////////
PARAM_TEST_CASE(HOG, Size, int)
{
cv::Size winSize;
Size winSize;
int type;
Mat img_rgb;
virtual void SetUp()
{
winSize = GET_PARAM(0);
type = GET_PARAM(1);
img_rgb = readImage(workdir + "../gpu/road.png");
if(img_rgb.empty())
{
std::cout << "Couldn't read road.png" << std::endl;
}
}
};
TEST_P(HOG, GetDescriptors)
{
// Load image
cv::Mat img_rgb = readImage(workdir + "lena.jpg");
ASSERT_FALSE(img_rgb.empty());
// Convert image
cv::Mat img;
Mat img;
switch (type)
{
case CV_8UC1:
cv::cvtColor(img_rgb, img, CV_BGR2GRAY);
cvtColor(img_rgb, img, CV_BGR2GRAY);
break;
case CV_8UC4:
default:
cv::cvtColor(img_rgb, img, CV_BGR2BGRA);
cvtColor(img_rgb, img, CV_BGR2BGRA);
break;
}
cv::ocl::oclMat d_img(img);
ocl::oclMat d_img(img);
// HOGs
cv::ocl::HOGDescriptor ocl_hog;
ocl::HOGDescriptor ocl_hog;
ocl_hog.gamma_correction = true;
cv::HOGDescriptor hog;
HOGDescriptor hog;
hog.gammaCorrection = true;
// Compute descriptor
cv::ocl::oclMat d_descriptors;
ocl::oclMat d_descriptors;
ocl_hog.getDescriptors(d_img, ocl_hog.win_size, d_descriptors, ocl_hog.DESCR_FORMAT_COL_BY_COL);
cv::Mat down_descriptors;
Mat down_descriptors;
d_descriptors.download(down_descriptors);
down_descriptors = down_descriptors.reshape(0, down_descriptors.cols * down_descriptors.rows);
@ -105,45 +112,34 @@ TEST_P(HOG, GetDescriptors)
hog.compute(img_rgb, descriptors, ocl_hog.win_size);
break;
}
cv::Mat cpu_descriptors(descriptors);
Mat cpu_descriptors(descriptors);
EXPECT_MAT_SIMILAR(down_descriptors, cpu_descriptors, 1e-2);
}
bool match_rect(cv::Rect r1, cv::Rect r2, int threshold)
{
return ((abs(r1.x - r2.x) < threshold) && (abs(r1.y - r2.y) < threshold) &&
(abs(r1.width - r2.width) < threshold) && (abs(r1.height - r2.height) < threshold));
}
TEST_P(HOG, Detect)
{
// Load image
cv::Mat img_rgb = readImage(workdir + "lena.jpg");
ASSERT_FALSE(img_rgb.empty());
// Convert image
cv::Mat img;
Mat img;
switch (type)
{
case CV_8UC1:
cv::cvtColor(img_rgb, img, CV_BGR2GRAY);
cvtColor(img_rgb, img, CV_BGR2GRAY);
break;
case CV_8UC4:
default:
cv::cvtColor(img_rgb, img, CV_BGR2BGRA);
cvtColor(img_rgb, img, CV_BGR2BGRA);
break;
}
cv::ocl::oclMat d_img(img);
ocl::oclMat d_img(img);
// HOGs
if ((winSize != cv::Size(48, 96)) && (winSize != cv::Size(64, 128)))
winSize = cv::Size(64, 128);
cv::ocl::HOGDescriptor ocl_hog(winSize);
if ((winSize != Size(48, 96)) && (winSize != Size(64, 128)))
winSize = Size(64, 128);
ocl::HOGDescriptor ocl_hog(winSize);
ocl_hog.gamma_correction = true;
cv::HOGDescriptor hog;
HOGDescriptor hog;
hog.winSize = winSize;
hog.gammaCorrection = true;
@ -165,88 +161,117 @@ TEST_P(HOG, Detect)
}
// OpenCL detection
std::vector<cv::Rect> d_found;
ocl_hog.detectMultiScale(d_img, d_found, 0, cv::Size(8, 8), cv::Size(0, 0), 1.05, 2);
std::vector<Rect> d_found;
ocl_hog.detectMultiScale(d_img, d_found, 0, Size(8, 8), Size(0, 0), 1.05, 6);
// CPU detection
std::vector<cv::Rect> found;
std::vector<Rect> found;
switch (type)
{
case CV_8UC1:
hog.detectMultiScale(img, found, 0, cv::Size(8, 8), cv::Size(0, 0), 1.05, 2);
hog.detectMultiScale(img, found, 0, Size(8, 8), Size(0, 0), 1.05, 6);
break;
case CV_8UC4:
default:
hog.detectMultiScale(img_rgb, found, 0, cv::Size(8, 8), cv::Size(0, 0), 1.05, 2);
hog.detectMultiScale(img_rgb, found, 0, Size(8, 8), Size(0, 0), 1.05, 6);
break;
}
// Ground-truth rectangular people window
cv::Rect win1_64x128(231, 190, 72, 144);
cv::Rect win2_64x128(621, 156, 97, 194);
cv::Rect win1_48x96(238, 198, 63, 126);
cv::Rect win2_48x96(619, 161, 92, 185);
cv::Rect win3_48x96(488, 136, 56, 112);
// Compare whether ground-truth windows are detected and compare the number of windows detected.
std::vector<int> d_comp(4);
std::vector<int> comp(4);
for(int i = 0; i < (int)d_comp.size(); i++)
{
d_comp[i] = 0;
comp[i] = 0;
}
int threshold = 10;
int val = 32;
d_comp[0] = (int)d_found.size();
comp[0] = (int)found.size();
if (winSize == cv::Size(48, 96))
{
for(int i = 0; i < (int)d_found.size(); i++)
{
if (match_rect(d_found[i], win1_48x96, threshold))
d_comp[1] = val;
if (match_rect(d_found[i], win2_48x96, threshold))
d_comp[2] = val;
if (match_rect(d_found[i], win3_48x96, threshold))
d_comp[3] = val;
}
for(int i = 0; i < (int)found.size(); i++)
{
if (match_rect(found[i], win1_48x96, threshold))
comp[1] = val;
if (match_rect(found[i], win2_48x96, threshold))
comp[2] = val;
if (match_rect(found[i], win3_48x96, threshold))
comp[3] = val;
}
}
else if (winSize == cv::Size(64, 128))
{
for(int i = 0; i < (int)d_found.size(); i++)
{
if (match_rect(d_found[i], win1_64x128, threshold))
d_comp[1] = val;
if (match_rect(d_found[i], win2_64x128, threshold))
d_comp[2] = val;
}
for(int i = 0; i < (int)found.size(); i++)
{
if (match_rect(found[i], win1_64x128, threshold))
comp[1] = val;
if (match_rect(found[i], win2_64x128, threshold))
comp[2] = val;
}
}
EXPECT_MAT_NEAR(cv::Mat(d_comp), cv::Mat(comp), 3);
EXPECT_LT(checkRectSimilarity(img.size(), found, d_found), 1.0);
}
INSTANTIATE_TEST_CASE_P(OCL_ObjDetect, HOG, testing::Combine(
testing::Values(cv::Size(64, 128), cv::Size(48, 96)),
testing::Values(Size(64, 128), Size(48, 96)),
testing::Values(MatType(CV_8UC1), MatType(CV_8UC4))));
///////////////////////////// Haar //////////////////////////////
IMPLEMENT_PARAM_CLASS(CascadeName, std::string);
CascadeName cascade_frontalface_alt(std::string("haarcascade_frontalface_alt.xml"));
CascadeName cascade_frontalface_alt2(std::string("haarcascade_frontalface_alt2.xml"));
struct getRect
{
Rect operator ()(const CvAvgComp &e) const
{
return e.rect;
}
};
PARAM_TEST_CASE(Haar, int, CascadeName)
{
ocl::OclCascadeClassifier cascade, nestedCascade;
CascadeClassifier cpucascade, cpunestedCascade;
int flags;
std::string cascadeName;
vector<Rect> faces, oclfaces;
Mat img;
ocl::oclMat d_img;
virtual void SetUp()
{
flags = GET_PARAM(0);
cascadeName = (workdir + "../../data/haarcascades/").append(GET_PARAM(1));
if( (!cascade.load( cascadeName )) || (!cpucascade.load(cascadeName)) )
{
std::cout << "ERROR: Could not load classifier cascade" << std::endl;
return;
}
img = readImage(workdir + "lena.jpg", IMREAD_GRAYSCALE);
if(img.empty())
{
std::cout << "Couldn't read lena.jpg" << std::endl;
return ;
}
equalizeHist(img, img);
d_img.upload(img);
}
};
TEST_P(Haar, FaceDetect)
{
MemStorage storage(cvCreateMemStorage(0));
CvSeq *_objects;
_objects = cascade.oclHaarDetectObjects(d_img, storage, 1.1, 3,
flags, Size(30, 30), Size(0, 0));
vector<CvAvgComp> vecAvgComp;
Seq<CvAvgComp>(_objects).copyTo(vecAvgComp);
oclfaces.resize(vecAvgComp.size());
std::transform(vecAvgComp.begin(), vecAvgComp.end(), oclfaces.begin(), getRect());
cpucascade.detectMultiScale(img, faces, 1.1, 3,
flags,
Size(30, 30), Size(0, 0));
EXPECT_LT(checkRectSimilarity(img.size(), faces, oclfaces), 1.0);
}
TEST_P(Haar, FaceDetectUseBuf)
{
ocl::OclCascadeClassifierBuf cascadebuf;
if(!cascadebuf.load(cascadeName))
{
std::cout << "ERROR: Could not load classifier cascade for FaceDetectUseBuf!" << std::endl;
return;
}
cascadebuf.detectMultiScale(d_img, oclfaces, 1.1, 3,
flags,
Size(30, 30), Size(0, 0));
cpucascade.detectMultiScale(img, faces, 1.1, 3,
flags,
Size(30, 30), Size(0, 0));
// intentionally run ocl facedetect again and check if it still works after the first run
cascadebuf.detectMultiScale(d_img, oclfaces, 1.1, 3,
flags,
Size(30, 30));
cascadebuf.release();
EXPECT_LT(checkRectSimilarity(img.size(), faces, oclfaces), 1.0);
}
INSTANTIATE_TEST_CASE_P(OCL_ObjDetect, Haar,
Combine(Values(CV_HAAR_SCALE_IMAGE, 0),
Values(cascade_frontalface_alt/*, cascade_frontalface_alt2*/)));
#endif //HAVE_OPENCL

42
modules/ocl/test/test_pyrdown.cpp → modules/ocl/test/test_pyramids.cpp
View File

@ -15,7 +15,6 @@
// Third party copyrights are property of their respective owners.
//
// @Authors
// Dachuan Zhao, dachuan@multicorewareinc.com
// Yao Wang yao@multicorewareinc.com
//
// Redistribution and use in source and binary forms, with or without modification,
@ -56,11 +55,12 @@ using namespace cvtest;
using namespace testing;
using namespace std;
PARAM_TEST_CASE(PyrDown, MatType, int)
PARAM_TEST_CASE(PyrBase, MatType, int)
{
int type;
int channels;
Mat dst_cpu;
oclMat gdst;
virtual void SetUp()
{
type = GET_PARAM(0);
@ -69,19 +69,19 @@ PARAM_TEST_CASE(PyrDown, MatType, int)
};
/////////////////////// PyrDown //////////////////////////
struct PyrDown : PyrBase {};
TEST_P(PyrDown, Mat)
{
for(int j = 0; j < LOOP_TIMES; j++)
{
cv::Size size(MWIDTH, MHEIGHT);
cv::RNG &rng = TS::ptr()->get_rng();
cv::Mat src = randomMat(rng, size, CV_MAKETYPE(type, channels), 0, 100, false);
Size size(MWIDTH, MHEIGHT);
Mat src = randomMat(size, CV_MAKETYPE(type, channels));
oclMat gsrc(src);
cv::ocl::oclMat gsrc(src), gdst;
cv::Mat dst_cpu;
cv::pyrDown(src, dst_cpu);
cv::ocl::pyrDown(gsrc, gdst);
pyrDown(src, dst_cpu);
pyrDown(gsrc, gdst);
EXPECT_MAT_NEAR(dst_cpu, Mat(gdst), type == CV_32F ? 1e-4f : 1.0f);
}
@ -90,5 +90,27 @@ TEST_P(PyrDown, Mat)
INSTANTIATE_TEST_CASE_P(OCL_ImgProc, PyrDown, Combine(
Values(CV_8U, CV_32F), Values(1, 3, 4)));
/////////////////////// PyrUp //////////////////////////
struct PyrUp : PyrBase {};
TEST_P(PyrUp, Accuracy)
{
for(int j = 0; j < LOOP_TIMES; j++)
{
Size size(MWIDTH, MHEIGHT);
Mat src = randomMat(size, CV_MAKETYPE(type, channels));
oclMat gsrc(src);
pyrUp(src, dst_cpu);
pyrUp(gsrc, gdst);
EXPECT_MAT_NEAR(dst_cpu, Mat(gdst), (type == CV_32F ? 1e-4f : 1.0));
}
}
INSTANTIATE_TEST_CASE_P(OCL_ImgProc, PyrUp, testing::Combine(
Values(CV_8U, CV_32F), Values(1, 3, 4)));
#endif // HAVE_OPENCL

91
modules/ocl/test/test_pyrup.cpp
View File

@ -1,91 +0,0 @@
/*M///////////////////////////////////////////////////////////////////////////////////////
//
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
//
// By downloading, copying, installing or using the software you agree to this license.
// If you do not agree to this license, do not download, install,
// copy or use the software.
//
//
// License Agreement
// For Open Source Computer Vision Library
//
// Copyright (C) 2010-2012, Multicoreware, Inc., all rights reserved.
// Copyright (C) 2010-2012, Advanced Micro Devices, Inc., all rights reserved.
// Third party copyrights are property of their respective owners.
//
// @Authors
// Zhang Chunpeng chunpeng@multicorewareinc.com
// Yao Wang yao@multicorewareinc.com
//
// Redistribution and use in source and binary forms, with or without modification,
// are permitted provided that the following conditions are met:
//
// * Redistribution's of source code must retain the above copyright notice,
// this list of conditions and the following disclaimer.
//
// * Redistribution's in binary form must reproduce the above copyright notice,
// this list of conditions and the following disclaimer in the documentation
// and/or other oclMaterials provided with the distribution.
//
// * The name of the copyright holders may not be used to endorse or promote products
// derived from this software without specific prior written permission.
//
// This software is provided by the copyright holders and contributors "as is" and
// any express or implied warranties, including, but not limited to, the implied
// warranties of merchantability and fitness for a particular purpose are disclaimed.
// In no event shall the Intel Corporation or contributors be liable for any direct,
// indirect, incidental, special, exemplary, or consequential damages
// (including, but not limited to, procurement of substitute goods or services;
// loss of use, data, or profits; or business interruption) however caused
// and on any theory of liability, whether in contract, strict liability,
// or tort (including negligence or otherwise) arising in any way out of
// the use of this software, even if advised of the possibility of such damage.
//
//M*/
#include "precomp.hpp"
#include "opencv2/core/core.hpp"
#ifdef HAVE_OPENCL
using namespace cv;
using namespace cvtest;
using namespace testing;
using namespace std;
PARAM_TEST_CASE(PyrUp, MatType, int)
{
int type;
int channels;
virtual void SetUp()
{
type = GET_PARAM(0);
channels = GET_PARAM(1);
}
};
TEST_P(PyrUp, Accuracy)
{
for(int j = 0; j < LOOP_TIMES; j++)
{
Size size(MWIDTH, MHEIGHT);
Mat src = randomMat(size, CV_MAKETYPE(type, channels));
Mat dst_gold;
pyrUp(src, dst_gold);
ocl::oclMat dst;
ocl::oclMat srcMat(src);
ocl::pyrUp(srcMat, dst);
EXPECT_MAT_NEAR(dst_gold, Mat(dst), (type == CV_32F ? 1e-4f : 1.0));
}
}
INSTANTIATE_TEST_CASE_P(OCL_ImgProc, PyrUp, testing::Combine(
Values(CV_8U, CV_32F), Values(1, 3, 4)));
#endif // HAVE_OPENCL

102
modules/ocl/test/utility.cpp
View File

@ -100,12 +100,6 @@ Mat randomMat(Size size, int type, double minVal, double maxVal)
return randomMat(TS::ptr()->get_rng(), size, type, minVal, maxVal, false);
}
/*
void showDiff(InputArray gold_, InputArray actual_, double eps)
{
@ -137,58 +131,7 @@ void showDiff(InputArray gold_, InputArray actual_, double eps)
}
*/
/*
bool supportFeature(const DeviceInfo& info, FeatureSet feature)
{
return TargetArchs::builtWith(feature) && info.supports(feature);
}
const vector<DeviceInfo>& devices()
{
static vector<DeviceInfo> devs;
static bool first = true;
if (first)
{
int deviceCount = getCudaEnabledDeviceCount();
devs.reserve(deviceCount);
for (int i = 0; i < deviceCount; ++i)
{
DeviceInfo info(i);
if (info.isCompatible())
devs.push_back(info);
}
first = false;
}
return devs;
}
vector<DeviceInfo> devices(FeatureSet feature)
{
const vector<DeviceInfo>& d = devices();
vector<DeviceInfo> devs_filtered;
if (TargetArchs::builtWith(feature))
{
devs_filtered.reserve(d.size());
for (size_t i = 0, size = d.size(); i < size; ++i)
{
const DeviceInfo& info = d[i];
if (info.supports(feature))
devs_filtered.push_back(info);
}
}
return devs_filtered;
}
*/
vector<MatType> types(int depth_start, int depth_end, int cn_start, int cn_end)
{
@ -264,3 +207,48 @@ void PrintTo(const Inverse &inverse, std::ostream *os)
(*os) << "direct";
}
double checkRectSimilarity(Size sz, std::vector<Rect>& ob1, std::vector<Rect>& ob2)
{
double final_test_result = 0.0;
size_t sz1 = ob1.size();
size_t sz2 = ob2.size();
if(sz1 != sz2)
{
return sz1 > sz2 ? (double)(sz1 - sz2) : (double)(sz2 - sz1);
}
else
{
if(sz1==0 && sz2==0)
return 0;
cv::Mat cpu_result(sz, CV_8UC1);
cpu_result.setTo(0);
for(vector<Rect>::const_iterator r = ob1.begin(); r != ob1.end(); r++)
{
cv::Mat cpu_result_roi(cpu_result, *r);
cpu_result_roi.setTo(1);
cpu_result.copyTo(cpu_result);
}
int cpu_area = cv::countNonZero(cpu_result > 0);
cv::Mat gpu_result(sz, CV_8UC1);
gpu_result.setTo(0);
for(vector<Rect>::const_iterator r2 = ob2.begin(); r2 != ob2.end(); r2++)
{
cv::Mat gpu_result_roi(gpu_result, *r2);
gpu_result_roi.setTo(1);
gpu_result.copyTo(gpu_result);
}
cv::Mat result_;
multiply(cpu_result, gpu_result, result_);
int result = cv::countNonZero(result_ > 0);
if(cpu_area!=0 && result!=0)
final_test_result = 1.0 - (double)result/(double)cpu_area;
else if(cpu_area==0 && result!=0)
final_test_result = -1;
}
return final_test_result;
}

11
modules/ocl/test/utility.hpp
View File

@ -55,13 +55,12 @@ cv::Mat randomMat(cv::Size size, int type, double minVal = 0.0, double maxVal =
void showDiff(cv::InputArray gold, cv::InputArray actual, double eps);
//! return true if device supports specified feature and gpu module was built with support the feature.
//bool supportFeature(const cv::gpu::DeviceInfo& info, cv::gpu::FeatureSet feature);
// This function test if gpu_rst matches cpu_rst.
// If the two vectors are not equal, it will return the difference in vector size
// Else it will return (total diff of each cpu and gpu rects covered pixels)/(total cpu rects covered pixels)
// The smaller, the better matched
double checkRectSimilarity(cv::Size sz, std::vector<cv::Rect>& ob1, std::vector<cv::Rect>& ob2);
//! return all devices compatible with current gpu module build.
//const std::vector<cv::ocl::DeviceInfo>& devices();
//! return all devices compatible with current gpu module build which support specified feature.
//std::vector<cv::ocl::DeviceInfo> devices(cv::gpu::FeatureSet feature);
//! read image from testdata folder.
cv::Mat readImage(const std::string &fileName, int flags = cv::IMREAD_COLOR);