generic-library/opencv
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Normalize line endings and whitespace

Browse Source
This commit is contained in:
OpenCV Buildbot
2012-10-17 11:12:04 +04:00
committed by Andrey Kamaev
parent 0442bca235
commit 81f826db2b
1511 changed files with 258678 additions and 258624 deletions
Download Patch File Download Diff File Expand all files Collapse all files

140
samples/gpu/CMakeLists.txt
View File

@@ -1,70 +1,70 @@
SET(OPENCV_GPU_SAMPLES_REQUIRED_DEPS opencv_core opencv_flann opencv_imgproc opencv_highgui
opencv_ml opencv_video opencv_objdetect opencv_features2d
opencv_calib3d opencv_legacy opencv_contrib opencv_gpu
opencv_nonfree)
ocv_check_dependencies(${OPENCV_GPU_SAMPLES_REQUIRED_DEPS})
if(BUILD_EXAMPLES AND OCV_DEPENDENCIES_FOUND)
set(project "gpu")
string(TOUPPER "${project}" project_upper)
project("${project}_samples")
ocv_include_modules(${OPENCV_GPU_SAMPLES_REQUIRED_DEPS})
ocv_include_directories(
"${OpenCV_SOURCE_DIR}/modules/gpu/src/nvidia"
"${OpenCV_SOURCE_DIR}/modules/gpu/src/nvidia/core"
)
if(HAVE_CUDA)
ocv_include_directories(${CUDA_INCLUDE_DIRS})
endif()
if(CMAKE_COMPILER_IS_GNUCXX AND NOT ENABLE_NOISY_WARNINGS)
set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Wno-unused-function")
endif()
# ---------------------------------------------
# Define executable targets
# ---------------------------------------------
MACRO(OPENCV_DEFINE_GPU_EXAMPLE name srcs)
set(the_target "example_${project}_${name}")
add_executable(${the_target} ${srcs})
target_link_libraries(${the_target} ${OPENCV_LINKER_LIBS} ${OPENCV_GPU_SAMPLES_REQUIRED_DEPS})
set_target_properties(${the_target} PROPERTIES
OUTPUT_NAME "${name}_${project}"
PROJECT_LABEL "(EXAMPLE_${project_upper}) ${name}")
if(ENABLE_SOLUTION_FOLDERS)
set_target_properties(${the_target} PROPERTIES FOLDER "samples//${project}")
endif()
if(WIN32)
if(MSVC AND NOT BUILD_SHARED_LIBS)
set_target_properties(${the_target} PROPERTIES LINK_FLAGS "/NODEFAULTLIB:atlthunk.lib /NODEFAULTLIB:atlsd.lib /DEBUG")
endif()
install(TARGETS ${the_target} RUNTIME DESTINATION "samples/${project}" COMPONENT main)
endif()
ENDMACRO()
file(GLOB all_samples RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} *.cpp)
foreach(sample_filename ${all_samples})
get_filename_component(sample ${sample_filename} NAME_WE)
file(GLOB sample_srcs RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} ${sample}.*)
OPENCV_DEFINE_GPU_EXAMPLE(${sample} ${sample_srcs})
endforeach()
include("performance/CMakeLists.txt")
endif()
if (INSTALL_C_EXAMPLES AND NOT WIN32)
file(GLOB install_list *.c *.cpp *.jpg *.png *.data makefile.* build_all.sh *.dsp *.cmd )
install(FILES ${install_list}
DESTINATION share/OpenCV/samples/${project}
PERMISSIONS OWNER_READ GROUP_READ WORLD_READ)
endif()
SET(OPENCV_GPU_SAMPLES_REQUIRED_DEPS opencv_core opencv_flann opencv_imgproc opencv_highgui
opencv_ml opencv_video opencv_objdetect opencv_features2d
opencv_calib3d opencv_legacy opencv_contrib opencv_gpu
opencv_nonfree)
ocv_check_dependencies(${OPENCV_GPU_SAMPLES_REQUIRED_DEPS})
if(BUILD_EXAMPLES AND OCV_DEPENDENCIES_FOUND)
set(project "gpu")
string(TOUPPER "${project}" project_upper)
project("${project}_samples")
ocv_include_modules(${OPENCV_GPU_SAMPLES_REQUIRED_DEPS})
ocv_include_directories(
"${OpenCV_SOURCE_DIR}/modules/gpu/src/nvidia"
"${OpenCV_SOURCE_DIR}/modules/gpu/src/nvidia/core"
)
if(HAVE_CUDA)
ocv_include_directories(${CUDA_INCLUDE_DIRS})
endif()
if(CMAKE_COMPILER_IS_GNUCXX AND NOT ENABLE_NOISY_WARNINGS)
set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Wno-unused-function")
endif()
# ---------------------------------------------
# Define executable targets
# ---------------------------------------------
MACRO(OPENCV_DEFINE_GPU_EXAMPLE name srcs)
set(the_target "example_${project}_${name}")
add_executable(${the_target} ${srcs})
target_link_libraries(${the_target} ${OPENCV_LINKER_LIBS} ${OPENCV_GPU_SAMPLES_REQUIRED_DEPS})
set_target_properties(${the_target} PROPERTIES
OUTPUT_NAME "${name}_${project}"
PROJECT_LABEL "(EXAMPLE_${project_upper}) ${name}")
if(ENABLE_SOLUTION_FOLDERS)
set_target_properties(${the_target} PROPERTIES FOLDER "samples//${project}")
endif()
if(WIN32)
if(MSVC AND NOT BUILD_SHARED_LIBS)
set_target_properties(${the_target} PROPERTIES LINK_FLAGS "/NODEFAULTLIB:atlthunk.lib /NODEFAULTLIB:atlsd.lib /DEBUG")
endif()
install(TARGETS ${the_target} RUNTIME DESTINATION "samples/${project}" COMPONENT main)
endif()
ENDMACRO()
file(GLOB all_samples RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} *.cpp)
foreach(sample_filename ${all_samples})
get_filename_component(sample ${sample_filename} NAME_WE)
file(GLOB sample_srcs RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} ${sample}.*)
OPENCV_DEFINE_GPU_EXAMPLE(${sample} ${sample_srcs})
endforeach()
include("performance/CMakeLists.txt")
endif()
if (INSTALL_C_EXAMPLES AND NOT WIN32)
file(GLOB install_list *.c *.cpp *.jpg *.png *.data makefile.* build_all.sh *.dsp *.cmd )
install(FILES ${install_list}
DESTINATION share/OpenCV/samples/${project}
PERMISSIONS OWNER_READ GROUP_READ WORLD_READ)
endif()

136
samples/gpu/alpha_comp.cpp
View File

@@ -1,68 +1,68 @@
#include <iostream>
#include "opencv2/core/opengl_interop.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/gpu/gpu.hpp"
using namespace std;
using namespace cv;
using namespace cv::gpu;
int main()
{
cout << "This program demonstrates using alphaComp" << endl;
cout << "Press SPACE to change compositing operation" << endl;
cout << "Press ESC to exit" << endl;
namedWindow("First Image", WINDOW_NORMAL);
namedWindow("Second Image", WINDOW_NORMAL);
namedWindow("Result", WINDOW_OPENGL);
setGlDevice();
Mat src1(640, 480, CV_8UC4, Scalar::all(0));
Mat src2(640, 480, CV_8UC4, Scalar::all(0));
rectangle(src1, Rect(50, 50, 200, 200), Scalar(0, 0, 255, 128), 30);
rectangle(src2, Rect(100, 100, 200, 200), Scalar(255, 0, 0, 128), 30);
GpuMat d_src1(src1);
GpuMat d_src2(src2);
GpuMat d_res;
imshow("First Image", src1);
imshow("Second Image", src2);
int alpha_op = ALPHA_OVER;
const char* op_names[] =
{
"ALPHA_OVER", "ALPHA_IN", "ALPHA_OUT", "ALPHA_ATOP", "ALPHA_XOR", "ALPHA_PLUS", "ALPHA_OVER_PREMUL", "ALPHA_IN_PREMUL", "ALPHA_OUT_PREMUL",
"ALPHA_ATOP_PREMUL", "ALPHA_XOR_PREMUL", "ALPHA_PLUS_PREMUL", "ALPHA_PREMUL"
};
for(;;)
{
cout << op_names[alpha_op] << endl;
alphaComp(d_src1, d_src2, d_res, alpha_op);
imshow("Result", d_res);
char key = static_cast<char>(waitKey());
if (key == 27)
break;
if (key == 32)
{
++alpha_op;
if (alpha_op > ALPHA_PREMUL)
alpha_op = ALPHA_OVER;
}
}
return 0;
}
#include <iostream>
#include "opencv2/core/opengl_interop.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/gpu/gpu.hpp"
using namespace std;
using namespace cv;
using namespace cv::gpu;
int main()
{
cout << "This program demonstrates using alphaComp" << endl;
cout << "Press SPACE to change compositing operation" << endl;
cout << "Press ESC to exit" << endl;
namedWindow("First Image", WINDOW_NORMAL);
namedWindow("Second Image", WINDOW_NORMAL);
namedWindow("Result", WINDOW_OPENGL);
setGlDevice();
Mat src1(640, 480, CV_8UC4, Scalar::all(0));
Mat src2(640, 480, CV_8UC4, Scalar::all(0));
rectangle(src1, Rect(50, 50, 200, 200), Scalar(0, 0, 255, 128), 30);
rectangle(src2, Rect(100, 100, 200, 200), Scalar(255, 0, 0, 128), 30);
GpuMat d_src1(src1);
GpuMat d_src2(src2);
GpuMat d_res;
imshow("First Image", src1);
imshow("Second Image", src2);
int alpha_op = ALPHA_OVER;
const char* op_names[] =
{
"ALPHA_OVER", "ALPHA_IN", "ALPHA_OUT", "ALPHA_ATOP", "ALPHA_XOR", "ALPHA_PLUS", "ALPHA_OVER_PREMUL", "ALPHA_IN_PREMUL", "ALPHA_OUT_PREMUL",
"ALPHA_ATOP_PREMUL", "ALPHA_XOR_PREMUL", "ALPHA_PLUS_PREMUL", "ALPHA_PREMUL"
};
for(;;)
{
cout << op_names[alpha_op] << endl;
alphaComp(d_src1, d_src2, d_res, alpha_op);
imshow("Result", d_res);
char key = static_cast<char>(waitKey());
if (key == 27)
break;
if (key == 32)
{
++alpha_op;
if (alpha_op > ALPHA_PREMUL)
alpha_op = ALPHA_OVER;
}
}
return 0;
}

624
samples/gpu/brox_optical_flow.cpp
View File

@@ -1,312 +1,312 @@
#include <iostream>
#include <iomanip>
#include <string>
#include "cvconfig.h"
#include "opencv2/core/core.hpp"
#include "opencv2/core/opengl_interop.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/gpu/gpu.hpp"
using namespace std;
using namespace cv;
using namespace cv::gpu;
void getFlowField(const Mat& u, const Mat& v, Mat& flowField);
#ifdef HAVE_OPENGL
void needleMapDraw(void* userdata);
#endif
int main(int argc, const char* argv[])
{
try
{
const char* keys =
"{ h help | | print help message }"
"{ l left | | specify left image }"
"{ r right | | specify right image }"
"{ s scale | 0.8 | set pyramid scale factor }"
"{ a alpha | 0.197 | set alpha }"
"{ g gamma | 50.0 | set gamma }"
"{ i inner | 10 | set number of inner iterations }"
"{ o outer | 77 | set number of outer iterations }"
"{ si solver | 10 | set number of basic solver iterations }"
"{ t time_step | 0.1 | set frame interpolation time step }";
CommandLineParser cmd(argc, argv, keys);
if (cmd.has("help") || !cmd.check())
{
cmd.printMessage();
cmd.printErrors();
return 0;
}
string frame0Name = cmd.get<string>("left");
string frame1Name = cmd.get<string>("right");
float scale = cmd.get<float>("scale");
float alpha = cmd.get<float>("alpha");
float gamma = cmd.get<float>("gamma");
int inner_iterations = cmd.get<int>("inner");
int outer_iterations = cmd.get<int>("outer");
int solver_iterations = cmd.get<int>("solver");
float timeStep = cmd.get<float>("time_step");
if (frame0Name.empty() || frame1Name.empty())
{
cerr << "Missing input file names" << endl;
return -1;
}
Mat frame0Color = imread(frame0Name);
Mat frame1Color = imread(frame1Name);
if (frame0Color.empty() || frame1Color.empty())
{
cout << "Can't load input images" << endl;
return -1;
}
cv::gpu::printShortCudaDeviceInfo(cv::gpu::getDevice());
cout << "OpenCV / NVIDIA Computer Vision" << endl;
cout << "Optical Flow Demo: Frame Interpolation" << endl;
cout << "=========================================" << endl;
namedWindow("Forward flow");
namedWindow("Backward flow");
namedWindow("Needle Map", WINDOW_OPENGL);
namedWindow("Interpolated frame");
setGlDevice();
cout << "Press:" << endl;
cout << "\tESC to quit" << endl;
cout << "\t'a' to move to the previous frame" << endl;
cout << "\t's' to move to the next frame\n" << endl;
frame0Color.convertTo(frame0Color, CV_32F, 1.0 / 255.0);
frame1Color.convertTo(frame1Color, CV_32F, 1.0 / 255.0);
Mat frame0Gray, frame1Gray;
cvtColor(frame0Color, frame0Gray, COLOR_BGR2GRAY);
cvtColor(frame1Color, frame1Gray, COLOR_BGR2GRAY);
GpuMat d_frame0(frame0Gray);
GpuMat d_frame1(frame1Gray);
cout << "Estimating optical flow" << endl;
BroxOpticalFlow d_flow(alpha, gamma, scale, inner_iterations, outer_iterations, solver_iterations);
cout << "\tForward..." << endl;
GpuMat d_fu, d_fv;
d_flow(d_frame0, d_frame1, d_fu, d_fv);
Mat flowFieldForward;
getFlowField(Mat(d_fu), Mat(d_fv), flowFieldForward);
cout << "\tBackward..." << endl;
GpuMat d_bu, d_bv;
d_flow(d_frame1, d_frame0, d_bu, d_bv);
Mat flowFieldBackward;
getFlowField(Mat(d_bu), Mat(d_bv), flowFieldBackward);
#ifdef HAVE_OPENGL
cout << "Create Optical Flow Needle Map..." << endl;
GpuMat d_vertex, d_colors;
createOpticalFlowNeedleMap(d_fu, d_fv, d_vertex, d_colors);
#endif
cout << "Interpolating..." << endl;
// first frame color components
GpuMat d_b, d_g, d_r;
// second frame color components
GpuMat d_bt, d_gt, d_rt;
// prepare color components on host and copy them to device memory
Mat channels[3];
cv::split(frame0Color, channels);
d_b.upload(channels[0]);
d_g.upload(channels[1]);
d_r.upload(channels[2]);
cv::split(frame1Color, channels);
d_bt.upload(channels[0]);
d_gt.upload(channels[1]);
d_rt.upload(channels[2]);
// temporary buffer
GpuMat d_buf;
// intermediate frame color components (GPU memory)
GpuMat d_rNew, d_gNew, d_bNew;
GpuMat d_newFrame;
vector<Mat> frames;
frames.reserve(static_cast<int>(1.0f / timeStep) + 2);
frames.push_back(frame0Color);
// compute interpolated frames
for (float timePos = timeStep; timePos < 1.0f; timePos += timeStep)
{
// interpolate blue channel
interpolateFrames(d_b, d_bt, d_fu, d_fv, d_bu, d_bv, timePos, d_bNew, d_buf);
// interpolate green channel
interpolateFrames(d_g, d_gt, d_fu, d_fv, d_bu, d_bv, timePos, d_gNew, d_buf);
// interpolate red channel
interpolateFrames(d_r, d_rt, d_fu, d_fv, d_bu, d_bv, timePos, d_rNew, d_buf);
GpuMat channels3[] = {d_bNew, d_gNew, d_rNew};
merge(channels3, 3, d_newFrame);
frames.push_back(Mat(d_newFrame));
cout << setprecision(4) << timePos * 100.0f << "%\r";
}
frames.push_back(frame1Color);
cout << setw(5) << "100%" << endl;
cout << "Done" << endl;
imshow("Forward flow", flowFieldForward);
imshow("Backward flow", flowFieldBackward);
#ifdef HAVE_OPENGL
GlArrays arr;
arr.setVertexArray(d_vertex);
arr.setColorArray(d_colors, false);
setOpenGlDrawCallback("Needle Map", needleMapDraw, &arr);
#endif
int currentFrame = 0;
imshow("Interpolated frame", frames[currentFrame]);
for(;;)
{
int key = toupper(waitKey(10) & 0xff);
switch (key)
{
case 27:
return 0;
case 'A':
if (currentFrame > 0)
--currentFrame;
imshow("Interpolated frame", frames[currentFrame]);
break;
case 'S':
if (currentFrame < static_cast<int>(frames.size()) - 1)
++currentFrame;
imshow("Interpolated frame", frames[currentFrame]);
break;
}
}
}
catch (const exception& ex)
{
cerr << ex.what() << endl;
return -1;
}
catch (...)
{
cerr << "Unknow error" << endl;
return -1;
}
}
template <typename T> inline T clamp (T x, T a, T b)
{
return ((x) > (a) ? ((x) < (b) ? (x) : (b)) : (a));
}
template <typename T> inline T mapValue(T x, T a, T b, T c, T d)
{
x = clamp(x, a, b);
return c + (d - c) * (x - a) / (b - a);
}
void getFlowField(const Mat& u, const Mat& v, Mat& flowField)
{
float maxDisplacement = 1.0f;
for (int i = 0; i < u.rows; ++i)
{
const float* ptr_u = u.ptr<float>(i);
const float* ptr_v = v.ptr<float>(i);
for (int j = 0; j < u.cols; ++j)
{
float d = max(fabsf(ptr_u[j]), fabsf(ptr_v[j]));
if (d > maxDisplacement)
maxDisplacement = d;
}
}
flowField.create(u.size(), CV_8UC4);
for (int i = 0; i < flowField.rows; ++i)
{
const float* ptr_u = u.ptr<float>(i);
const float* ptr_v = v.ptr<float>(i);
Vec4b* row = flowField.ptr<Vec4b>(i);
for (int j = 0; j < flowField.cols; ++j)
{
row[j][0] = 0;
row[j][1] = static_cast<unsigned char> (mapValue (-ptr_v[j], -maxDisplacement, maxDisplacement, 0.0f, 255.0f));
row[j][2] = static_cast<unsigned char> (mapValue ( ptr_u[j], -maxDisplacement, maxDisplacement, 0.0f, 255.0f));
row[j][3] = 255;
}
}
}
#ifdef HAVE_OPENGL
void needleMapDraw(void* userdata)
{
const GlArrays* arr = static_cast<const GlArrays*>(userdata);
GlCamera camera;
camera.setOrthoProjection(0.0, 1.0, 1.0, 0.0, 0.0, 1.0);
camera.lookAt(Point3d(0.0, 0.0, 1.0), Point3d(0.0, 0.0, 0.0), Point3d(0.0, 1.0, 0.0));
camera.setupProjectionMatrix();
camera.setupModelViewMatrix();
render(*arr, RenderMode::TRIANGLES);
}
#endif
#include <iostream>
#include <iomanip>
#include <string>
#include "cvconfig.h"
#include "opencv2/core/core.hpp"
#include "opencv2/core/opengl_interop.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/gpu/gpu.hpp"
using namespace std;
using namespace cv;
using namespace cv::gpu;
void getFlowField(const Mat& u, const Mat& v, Mat& flowField);
#ifdef HAVE_OPENGL
void needleMapDraw(void* userdata);
#endif
int main(int argc, const char* argv[])
{
try
{
const char* keys =
"{ h help | | print help message }"
"{ l left | | specify left image }"
"{ r right | | specify right image }"
"{ s scale | 0.8 | set pyramid scale factor }"
"{ a alpha | 0.197 | set alpha }"
"{ g gamma | 50.0 | set gamma }"
"{ i inner | 10 | set number of inner iterations }"
"{ o outer | 77 | set number of outer iterations }"
"{ si solver | 10 | set number of basic solver iterations }"
"{ t time_step | 0.1 | set frame interpolation time step }";
CommandLineParser cmd(argc, argv, keys);
if (cmd.has("help") || !cmd.check())
{
cmd.printMessage();
cmd.printErrors();
return 0;
}
string frame0Name = cmd.get<string>("left");
string frame1Name = cmd.get<string>("right");
float scale = cmd.get<float>("scale");
float alpha = cmd.get<float>("alpha");
float gamma = cmd.get<float>("gamma");
int inner_iterations = cmd.get<int>("inner");
int outer_iterations = cmd.get<int>("outer");
int solver_iterations = cmd.get<int>("solver");
float timeStep = cmd.get<float>("time_step");
if (frame0Name.empty() || frame1Name.empty())
{
cerr << "Missing input file names" << endl;
return -1;
}
Mat frame0Color = imread(frame0Name);
Mat frame1Color = imread(frame1Name);
if (frame0Color.empty() || frame1Color.empty())
{
cout << "Can't load input images" << endl;
return -1;
}
cv::gpu::printShortCudaDeviceInfo(cv::gpu::getDevice());
cout << "OpenCV / NVIDIA Computer Vision" << endl;
cout << "Optical Flow Demo: Frame Interpolation" << endl;
cout << "=========================================" << endl;
namedWindow("Forward flow");
namedWindow("Backward flow");
namedWindow("Needle Map", WINDOW_OPENGL);
namedWindow("Interpolated frame");
setGlDevice();
cout << "Press:" << endl;
cout << "\tESC to quit" << endl;
cout << "\t'a' to move to the previous frame" << endl;
cout << "\t's' to move to the next frame\n" << endl;
frame0Color.convertTo(frame0Color, CV_32F, 1.0 / 255.0);
frame1Color.convertTo(frame1Color, CV_32F, 1.0 / 255.0);
Mat frame0Gray, frame1Gray;
cvtColor(frame0Color, frame0Gray, COLOR_BGR2GRAY);
cvtColor(frame1Color, frame1Gray, COLOR_BGR2GRAY);
GpuMat d_frame0(frame0Gray);
GpuMat d_frame1(frame1Gray);
cout << "Estimating optical flow" << endl;
BroxOpticalFlow d_flow(alpha, gamma, scale, inner_iterations, outer_iterations, solver_iterations);
cout << "\tForward..." << endl;
GpuMat d_fu, d_fv;
d_flow(d_frame0, d_frame1, d_fu, d_fv);
Mat flowFieldForward;
getFlowField(Mat(d_fu), Mat(d_fv), flowFieldForward);
cout << "\tBackward..." << endl;
GpuMat d_bu, d_bv;
d_flow(d_frame1, d_frame0, d_bu, d_bv);
Mat flowFieldBackward;
getFlowField(Mat(d_bu), Mat(d_bv), flowFieldBackward);
#ifdef HAVE_OPENGL
cout << "Create Optical Flow Needle Map..." << endl;
GpuMat d_vertex, d_colors;
createOpticalFlowNeedleMap(d_fu, d_fv, d_vertex, d_colors);
#endif
cout << "Interpolating..." << endl;
// first frame color components
GpuMat d_b, d_g, d_r;
// second frame color components
GpuMat d_bt, d_gt, d_rt;
// prepare color components on host and copy them to device memory
Mat channels[3];
cv::split(frame0Color, channels);
d_b.upload(channels[0]);
d_g.upload(channels[1]);
d_r.upload(channels[2]);
cv::split(frame1Color, channels);
d_bt.upload(channels[0]);
d_gt.upload(channels[1]);
d_rt.upload(channels[2]);
// temporary buffer
GpuMat d_buf;
// intermediate frame color components (GPU memory)
GpuMat d_rNew, d_gNew, d_bNew;
GpuMat d_newFrame;
vector<Mat> frames;
frames.reserve(static_cast<int>(1.0f / timeStep) + 2);
frames.push_back(frame0Color);
// compute interpolated frames
for (float timePos = timeStep; timePos < 1.0f; timePos += timeStep)
{
// interpolate blue channel
interpolateFrames(d_b, d_bt, d_fu, d_fv, d_bu, d_bv, timePos, d_bNew, d_buf);
// interpolate green channel
interpolateFrames(d_g, d_gt, d_fu, d_fv, d_bu, d_bv, timePos, d_gNew, d_buf);
// interpolate red channel
interpolateFrames(d_r, d_rt, d_fu, d_fv, d_bu, d_bv, timePos, d_rNew, d_buf);
GpuMat channels3[] = {d_bNew, d_gNew, d_rNew};
merge(channels3, 3, d_newFrame);
frames.push_back(Mat(d_newFrame));
cout << setprecision(4) << timePos * 100.0f << "%\r";
}
frames.push_back(frame1Color);
cout << setw(5) << "100%" << endl;
cout << "Done" << endl;
imshow("Forward flow", flowFieldForward);
imshow("Backward flow", flowFieldBackward);
#ifdef HAVE_OPENGL
GlArrays arr;
arr.setVertexArray(d_vertex);
arr.setColorArray(d_colors, false);
setOpenGlDrawCallback("Needle Map", needleMapDraw, &arr);
#endif
int currentFrame = 0;
imshow("Interpolated frame", frames[currentFrame]);
for(;;)
{
int key = toupper(waitKey(10) & 0xff);
switch (key)
{
case 27:
return 0;
case 'A':
if (currentFrame > 0)
--currentFrame;
imshow("Interpolated frame", frames[currentFrame]);
break;
case 'S':
if (currentFrame < static_cast<int>(frames.size()) - 1)
++currentFrame;
imshow("Interpolated frame", frames[currentFrame]);
break;
}
}
}
catch (const exception& ex)
{
cerr << ex.what() << endl;
return -1;
}
catch (...)
{
cerr << "Unknow error" << endl;
return -1;
}
}
template <typename T> inline T clamp (T x, T a, T b)
{
return ((x) > (a) ? ((x) < (b) ? (x) : (b)) : (a));
}
template <typename T> inline T mapValue(T x, T a, T b, T c, T d)
{
x = clamp(x, a, b);
return c + (d - c) * (x - a) / (b - a);
}
void getFlowField(const Mat& u, const Mat& v, Mat& flowField)
{
float maxDisplacement = 1.0f;
for (int i = 0; i < u.rows; ++i)
{
const float* ptr_u = u.ptr<float>(i);
const float* ptr_v = v.ptr<float>(i);
for (int j = 0; j < u.cols; ++j)
{
float d = max(fabsf(ptr_u[j]), fabsf(ptr_v[j]));
if (d > maxDisplacement)
maxDisplacement = d;
}
}
flowField.create(u.size(), CV_8UC4);
for (int i = 0; i < flowField.rows; ++i)
{
const float* ptr_u = u.ptr<float>(i);
const float* ptr_v = v.ptr<float>(i);
Vec4b* row = flowField.ptr<Vec4b>(i);
for (int j = 0; j < flowField.cols; ++j)
{
row[j][0] = 0;
row[j][1] = static_cast<unsigned char> (mapValue (-ptr_v[j], -maxDisplacement, maxDisplacement, 0.0f, 255.0f));
row[j][2] = static_cast<unsigned char> (mapValue ( ptr_u[j], -maxDisplacement, maxDisplacement, 0.0f, 255.0f));
row[j][3] = 255;
}
}
}
#ifdef HAVE_OPENGL
void needleMapDraw(void* userdata)
{
const GlArrays* arr = static_cast<const GlArrays*>(userdata);
GlCamera camera;
camera.setOrthoProjection(0.0, 1.0, 1.0, 0.0, 0.0, 1.0);
camera.lookAt(Point3d(0.0, 0.0, 1.0), Point3d(0.0, 0.0, 0.0), Point3d(0.0, 1.0, 0.0));
camera.setupProjectionMatrix();
camera.setupModelViewMatrix();
render(*arr, RenderMode::TRIANGLES);
}
#endif

616
samples/gpu/cascadeclassifier.cpp
View File

@@ -1,308 +1,308 @@
// WARNING: this sample is under construction! Use it on your own risk.
#if defined _MSC_VER && _MSC_VER >= 1400
#pragma warning(disable : 4100)
#endif
#include <iostream>
#include <iomanip>
#include "opencv2/contrib/contrib.hpp"
#include "opencv2/objdetect/objdetect.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/imgproc/imgproc.hpp"
#include "opencv2/gpu/gpu.hpp"
using namespace std;
using namespace cv;
using namespace cv::gpu;
static void help()
{
cout << "Usage: ./cascadeclassifier_gpu \n\t--cascade <cascade_file>\n\t(<image>|--video <video>|--camera <camera_id>)\n"
"Using OpenCV version " << CV_VERSION << endl << endl;
}
template<class T>
void convertAndResize(const T& src, T& gray, T& resized, double scale)
{
if (src.channels() == 3)
{
cvtColor( src, gray, CV_BGR2GRAY );
}
else
{
gray = src;
}
Size sz(cvRound(gray.cols * scale), cvRound(gray.rows * scale));
if (scale != 1)
{
resize(gray, resized, sz);
}
else
{
resized = gray;
}
}
static void matPrint(Mat &img, int lineOffsY, Scalar fontColor, const string &ss)
{
int fontFace = FONT_HERSHEY_DUPLEX;
double fontScale = 0.8;
int fontThickness = 2;
Size fontSize = cv::getTextSize("T[]", fontFace, fontScale, fontThickness, 0);
Point org;
org.x = 1;
org.y = 3 * fontSize.height * (lineOffsY + 1) / 2;
putText(img, ss, org, fontFace, fontScale, CV_RGB(0,0,0), 5*fontThickness/2, 16);
putText(img, ss, org, fontFace, fontScale, fontColor, fontThickness, 16);
}
static void displayState(Mat &canvas, bool bHelp, bool bGpu, bool bLargestFace, bool bFilter, double fps)
{
Scalar fontColorRed = CV_RGB(255,0,0);
Scalar fontColorNV = CV_RGB(118,185,0);
ostringstream ss;
ss << "FPS = " << setprecision(1) << fixed << fps;
matPrint(canvas, 0, fontColorRed, ss.str());
ss.str("");
ss << "[" << canvas.cols << "x" << canvas.rows << "], " <<
(bGpu ? "GPU, " : "CPU, ") <<
(bLargestFace ? "OneFace, " : "MultiFace, ") <<
(bFilter ? "Filter:ON" : "Filter:OFF");
matPrint(canvas, 1, fontColorRed, ss.str());
// by Anatoly. MacOS fix. ostringstream(const string&) is a private
// matPrint(canvas, 2, fontColorNV, ostringstream("Space - switch GPU / CPU"));
if (bHelp)
{
matPrint(canvas, 2, fontColorNV, "Space - switch GPU / CPU");
matPrint(canvas, 3, fontColorNV, "M - switch OneFace / MultiFace");
matPrint(canvas, 4, fontColorNV, "F - toggle rectangles Filter");
matPrint(canvas, 5, fontColorNV, "H - toggle hotkeys help");
matPrint(canvas, 6, fontColorNV, "1/Q - increase/decrease scale");
}
else
{
matPrint(canvas, 2, fontColorNV, "H - toggle hotkeys help");
}
}
int main(int argc, const char *argv[])
{
if (argc == 1)
{
help();
return -1;
}
if (getCudaEnabledDeviceCount() == 0)
{
return cerr << "No GPU found or the library is compiled without GPU support" << endl, -1;
}
cv::gpu::printShortCudaDeviceInfo(cv::gpu::getDevice());
string cascadeName;
string inputName;
bool isInputImage = false;
bool isInputVideo = false;
bool isInputCamera = false;
for (int i = 1; i < argc; ++i)
{
if (string(argv[i]) == "--cascade")
cascadeName = argv[++i];
else if (string(argv[i]) == "--video")
{
inputName = argv[++i];
isInputVideo = true;
}
else if (string(argv[i]) == "--camera")
{
inputName = argv[++i];
isInputCamera = true;
}
else if (string(argv[i]) == "--help")
{
help();
return -1;
}
else if (!isInputImage)
{
inputName = argv[i];
isInputImage = true;
}
else
{
cout << "Unknown key: " << argv[i] << endl;
return -1;
}
}
CascadeClassifier_GPU cascade_gpu;
if (!cascade_gpu.load(cascadeName))
{
return cerr << "ERROR: Could not load cascade classifier \"" << cascadeName << "\"" << endl, help(), -1;
}
CascadeClassifier cascade_cpu;
if (!cascade_cpu.load(cascadeName))
{
return cerr << "ERROR: Could not load cascade classifier \"" << cascadeName << "\"" << endl, help(), -1;
}
VideoCapture capture;
Mat image;
if (isInputImage)
{
image = imread(inputName);
CV_Assert(!image.empty());
}
else if (isInputVideo)
{
capture.open(inputName);
CV_Assert(capture.isOpened());
}
else
{
capture.open(atoi(inputName.c_str()));
CV_Assert(capture.isOpened());
}
namedWindow("result", 1);
Mat frame, frame_cpu, gray_cpu, resized_cpu, faces_downloaded, frameDisp;
vector<Rect> facesBuf_cpu;
GpuMat frame_gpu, gray_gpu, resized_gpu, facesBuf_gpu;
/* parameters */
bool useGPU = true;
double scaleFactor = 1.0;
bool findLargestObject = false;
bool filterRects = true;
bool helpScreen = false;
int detections_num;
for (;;)
{
if (isInputCamera || isInputVideo)
{
capture >> frame;
if (frame.empty())
{
break;
}
}
(image.empty() ? frame : image).copyTo(frame_cpu);
frame_gpu.upload(image.empty() ? frame : image);
convertAndResize(frame_gpu, gray_gpu, resized_gpu, scaleFactor);
convertAndResize(frame_cpu, gray_cpu, resized_cpu, scaleFactor);
TickMeter tm;
tm.start();
if (useGPU)
{
cascade_gpu.visualizeInPlace = true;
cascade_gpu.findLargestObject = findLargestObject;
detections_num = cascade_gpu.detectMultiScale(resized_gpu, facesBuf_gpu, 1.2,
(filterRects || findLargestObject) ? 4 : 0);
facesBuf_gpu.colRange(0, detections_num).download(faces_downloaded);
}
else
{
Size minSize = cascade_gpu.getClassifierSize();
cascade_cpu.detectMultiScale(resized_cpu, facesBuf_cpu, 1.2,
(filterRects || findLargestObject) ? 4 : 0,
(findLargestObject ? CV_HAAR_FIND_BIGGEST_OBJECT : 0)
| CV_HAAR_SCALE_IMAGE,
minSize);
detections_num = (int)facesBuf_cpu.size();
}
if (!useGPU && detections_num)
{
for (int i = 0; i < detections_num; ++i)
{
rectangle(resized_cpu, facesBuf_cpu[i], Scalar(255));
}
}
if (useGPU)
{
resized_gpu.download(resized_cpu);
}
tm.stop();
double detectionTime = tm.getTimeMilli();
double fps = 1000 / detectionTime;
//print detections to console
cout << setfill(' ') << setprecision(2);
cout << setw(6) << fixed << fps << " FPS, " << detections_num << " det";
if ((filterRects || findLargestObject) && detections_num > 0)
{
Rect *faceRects = useGPU ? faces_downloaded.ptr<Rect>() : &facesBuf_cpu[0];
for (int i = 0; i < min(detections_num, 2); ++i)
{
cout << ", [" << setw(4) << faceRects[i].x
<< ", " << setw(4) << faceRects[i].y
<< ", " << setw(4) << faceRects[i].width
<< ", " << setw(4) << faceRects[i].height << "]";
}
}
cout << endl;
cvtColor(resized_cpu, frameDisp, CV_GRAY2BGR);
displayState(frameDisp, helpScreen, useGPU, findLargestObject, filterRects, fps);
imshow("result", frameDisp);
char key = (char)waitKey(5);
if (key == 27)
{
break;
}
switch (key)
{
case ' ':
useGPU = !useGPU;
break;
case 'm':
case 'M':
findLargestObject = !findLargestObject;
break;
case 'f':
case 'F':
filterRects = !filterRects;
break;
case '1':
scaleFactor *= 1.05;
break;
case 'q':
case 'Q':
scaleFactor /= 1.05;
break;
case 'h':
case 'H':
helpScreen = !helpScreen;
break;
}
}
return 0;
}
// WARNING: this sample is under construction! Use it on your own risk.
#if defined _MSC_VER && _MSC_VER >= 1400
#pragma warning(disable : 4100)
#endif
#include <iostream>
#include <iomanip>
#include "opencv2/contrib/contrib.hpp"
#include "opencv2/objdetect/objdetect.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/imgproc/imgproc.hpp"
#include "opencv2/gpu/gpu.hpp"
using namespace std;
using namespace cv;
using namespace cv::gpu;
static void help()
{
cout << "Usage: ./cascadeclassifier_gpu \n\t--cascade <cascade_file>\n\t(<image>|--video <video>|--camera <camera_id>)\n"
"Using OpenCV version " << CV_VERSION << endl << endl;
}
template<class T>
void convertAndResize(const T& src, T& gray, T& resized, double scale)
{
if (src.channels() == 3)
{
cvtColor( src, gray, CV_BGR2GRAY );
}
else
{
gray = src;
}
Size sz(cvRound(gray.cols * scale), cvRound(gray.rows * scale));
if (scale != 1)
{
resize(gray, resized, sz);
}
else
{
resized = gray;
}
}
static void matPrint(Mat &img, int lineOffsY, Scalar fontColor, const string &ss)
{
int fontFace = FONT_HERSHEY_DUPLEX;
double fontScale = 0.8;
int fontThickness = 2;
Size fontSize = cv::getTextSize("T[]", fontFace, fontScale, fontThickness, 0);
Point org;
org.x = 1;
org.y = 3 * fontSize.height * (lineOffsY + 1) / 2;
putText(img, ss, org, fontFace, fontScale, CV_RGB(0,0,0), 5*fontThickness/2, 16);
putText(img, ss, org, fontFace, fontScale, fontColor, fontThickness, 16);
}
static void displayState(Mat &canvas, bool bHelp, bool bGpu, bool bLargestFace, bool bFilter, double fps)
{
Scalar fontColorRed = CV_RGB(255,0,0);
Scalar fontColorNV = CV_RGB(118,185,0);
ostringstream ss;
ss << "FPS = " << setprecision(1) << fixed << fps;
matPrint(canvas, 0, fontColorRed, ss.str());
ss.str("");
ss << "[" << canvas.cols << "x" << canvas.rows << "], " <<
(bGpu ? "GPU, " : "CPU, ") <<
(bLargestFace ? "OneFace, " : "MultiFace, ") <<
(bFilter ? "Filter:ON" : "Filter:OFF");
matPrint(canvas, 1, fontColorRed, ss.str());
// by Anatoly. MacOS fix. ostringstream(const string&) is a private
// matPrint(canvas, 2, fontColorNV, ostringstream("Space - switch GPU / CPU"));
if (bHelp)
{
matPrint(canvas, 2, fontColorNV, "Space - switch GPU / CPU");
matPrint(canvas, 3, fontColorNV, "M - switch OneFace / MultiFace");
matPrint(canvas, 4, fontColorNV, "F - toggle rectangles Filter");
matPrint(canvas, 5, fontColorNV, "H - toggle hotkeys help");
matPrint(canvas, 6, fontColorNV, "1/Q - increase/decrease scale");
}
else
{
matPrint(canvas, 2, fontColorNV, "H - toggle hotkeys help");
}
}
int main(int argc, const char *argv[])
{
if (argc == 1)
{
help();
return -1;
}
if (getCudaEnabledDeviceCount() == 0)
{
return cerr << "No GPU found or the library is compiled without GPU support" << endl, -1;
}
cv::gpu::printShortCudaDeviceInfo(cv::gpu::getDevice());
string cascadeName;
string inputName;
bool isInputImage = false;
bool isInputVideo = false;
bool isInputCamera = false;
for (int i = 1; i < argc; ++i)
{
if (string(argv[i]) == "--cascade")
cascadeName = argv[++i];
else if (string(argv[i]) == "--video")
{
inputName = argv[++i];
isInputVideo = true;
}
else if (string(argv[i]) == "--camera")
{
inputName = argv[++i];
isInputCamera = true;
}
else if (string(argv[i]) == "--help")
{
help();
return -1;
}
else if (!isInputImage)
{
inputName = argv[i];
isInputImage = true;
}
else
{
cout << "Unknown key: " << argv[i] << endl;
return -1;
}
}
CascadeClassifier_GPU cascade_gpu;
if (!cascade_gpu.load(cascadeName))
{
return cerr << "ERROR: Could not load cascade classifier \"" << cascadeName << "\"" << endl, help(), -1;
}
CascadeClassifier cascade_cpu;
if (!cascade_cpu.load(cascadeName))
{
return cerr << "ERROR: Could not load cascade classifier \"" << cascadeName << "\"" << endl, help(), -1;
}
VideoCapture capture;
Mat image;
if (isInputImage)
{
image = imread(inputName);
CV_Assert(!image.empty());
}
else if (isInputVideo)
{
capture.open(inputName);
CV_Assert(capture.isOpened());
}
else
{
capture.open(atoi(inputName.c_str()));
CV_Assert(capture.isOpened());
}
namedWindow("result", 1);
Mat frame, frame_cpu, gray_cpu, resized_cpu, faces_downloaded, frameDisp;
vector<Rect> facesBuf_cpu;
GpuMat frame_gpu, gray_gpu, resized_gpu, facesBuf_gpu;
/* parameters */
bool useGPU = true;
double scaleFactor = 1.0;
bool findLargestObject = false;
bool filterRects = true;
bool helpScreen = false;
int detections_num;
for (;;)
{
if (isInputCamera || isInputVideo)
{
capture >> frame;
if (frame.empty())
{
break;
}
}
(image.empty() ? frame : image).copyTo(frame_cpu);
frame_gpu.upload(image.empty() ? frame : image);
convertAndResize(frame_gpu, gray_gpu, resized_gpu, scaleFactor);
convertAndResize(frame_cpu, gray_cpu, resized_cpu, scaleFactor);
TickMeter tm;
tm.start();
if (useGPU)
{
cascade_gpu.visualizeInPlace = true;
cascade_gpu.findLargestObject = findLargestObject;
detections_num = cascade_gpu.detectMultiScale(resized_gpu, facesBuf_gpu, 1.2,
(filterRects || findLargestObject) ? 4 : 0);
facesBuf_gpu.colRange(0, detections_num).download(faces_downloaded);
}
else
{
Size minSize = cascade_gpu.getClassifierSize();
cascade_cpu.detectMultiScale(resized_cpu, facesBuf_cpu, 1.2,
(filterRects || findLargestObject) ? 4 : 0,
(findLargestObject ? CV_HAAR_FIND_BIGGEST_OBJECT : 0)
| CV_HAAR_SCALE_IMAGE,
minSize);
detections_num = (int)facesBuf_cpu.size();
}
if (!useGPU && detections_num)
{
for (int i = 0; i < detections_num; ++i)
{
rectangle(resized_cpu, facesBuf_cpu[i], Scalar(255));
}
}
if (useGPU)
{
resized_gpu.download(resized_cpu);
}
tm.stop();
double detectionTime = tm.getTimeMilli();
double fps = 1000 / detectionTime;
//print detections to console
cout << setfill(' ') << setprecision(2);
cout << setw(6) << fixed << fps << " FPS, " << detections_num << " det";
if ((filterRects || findLargestObject) && detections_num > 0)
{
Rect *faceRects = useGPU ? faces_downloaded.ptr<Rect>() : &facesBuf_cpu[0];
for (int i = 0; i < min(detections_num, 2); ++i)
{
cout << ", [" << setw(4) << faceRects[i].x
<< ", " << setw(4) << faceRects[i].y
<< ", " << setw(4) << faceRects[i].width
<< ", " << setw(4) << faceRects[i].height << "]";
}
}
cout << endl;
cvtColor(resized_cpu, frameDisp, CV_GRAY2BGR);
displayState(frameDisp, helpScreen, useGPU, findLargestObject, filterRects, fps);
imshow("result", frameDisp);
char key = (char)waitKey(5);
if (key == 27)
{
break;
}
switch (key)
{
case ' ':
useGPU = !useGPU;
break;
case 'm':
case 'M':
findLargestObject = !findLargestObject;
break;
case 'f':
case 'F':
filterRects = !filterRects;
break;
case '1':
scaleFactor *= 1.05;
break;
case 'q':
case 'Q':
scaleFactor /= 1.05;
break;
case 'h':
case 'H':
helpScreen = !helpScreen;
break;
}
}
return 0;
}

760
samples/gpu/cascadeclassifier_nvidia_api.cpp
View File

@@ -1,380 +1,380 @@
#if defined _MSC_VER && _MSC_VER >= 1400
#pragma warning( disable : 4201 4408 4127 4100)
#endif
#include "cvconfig.h"
#include <iostream>
#include <iomanip>
#include <cstdio>
#include "opencv2/gpu/gpu.hpp"
#include "opencv2/highgui/highgui.hpp"
#ifdef HAVE_CUDA
#include "NCVHaarObjectDetection.hpp"
#endif
using namespace std;
using namespace cv;
#if !defined(HAVE_CUDA)
int main( int, const char** )
{
cout << "Please compile the library with CUDA support" << endl;
return -1;
}
#else
const Size2i preferredVideoFrameSize(640, 480);
const string wndTitle = "NVIDIA Computer Vision :: Haar Classifiers Cascade";
void matPrint(Mat &img, int lineOffsY, Scalar fontColor, const string &ss)
{
int fontFace = FONT_HERSHEY_DUPLEX;
double fontScale = 0.8;
int fontThickness = 2;
Size fontSize = cv::getTextSize("T[]", fontFace, fontScale, fontThickness, 0);
Point org;
org.x = 1;
org.y = 3 * fontSize.height * (lineOffsY + 1) / 2;
putText(img, ss, org, fontFace, fontScale, CV_RGB(0,0,0), 5*fontThickness/2, 16);
putText(img, ss, org, fontFace, fontScale, fontColor, fontThickness, 16);
}
void displayState(Mat &canvas, bool bHelp, bool bGpu, bool bLargestFace, bool bFilter, double fps)
{
Scalar fontColorRed = CV_RGB(255,0,0);
Scalar fontColorNV = CV_RGB(118,185,0);
ostringstream ss;
ss << "FPS = " << setprecision(1) << fixed << fps;
matPrint(canvas, 0, fontColorRed, ss.str());
ss.str("");
ss << "[" << canvas.cols << "x" << canvas.rows << "], " <<
(bGpu ? "GPU, " : "CPU, ") <<
(bLargestFace ? "OneFace, " : "MultiFace, ") <<
(bFilter ? "Filter:ON" : "Filter:OFF");
matPrint(canvas, 1, fontColorRed, ss.str());
if (bHelp)
{
matPrint(canvas, 2, fontColorNV, "Space - switch GPU / CPU");
matPrint(canvas, 3, fontColorNV, "M - switch OneFace / MultiFace");
matPrint(canvas, 4, fontColorNV, "F - toggle rectangles Filter");
matPrint(canvas, 5, fontColorNV, "H - toggle hotkeys help");
}
else
{
matPrint(canvas, 2, fontColorNV, "H - toggle hotkeys help");
}
}
NCVStatus process(Mat *srcdst,
Ncv32u width, Ncv32u height,
NcvBool bFilterRects, NcvBool bLargestFace,
HaarClassifierCascadeDescriptor &haar,
NCVVector<HaarStage64> &d_haarStages, NCVVector<HaarClassifierNode128> &d_haarNodes,
NCVVector<HaarFeature64> &d_haarFeatures, NCVVector<HaarStage64> &h_haarStages,
INCVMemAllocator &gpuAllocator,
INCVMemAllocator &cpuAllocator,
cudaDeviceProp &devProp)
{
ncvAssertReturn(!((srcdst == NULL) ^ gpuAllocator.isCounting()), NCV_NULL_PTR);
NCVStatus ncvStat;
NCV_SET_SKIP_COND(gpuAllocator.isCounting());
NCVMatrixAlloc<Ncv8u> d_src(gpuAllocator, width, height);
ncvAssertReturn(d_src.isMemAllocated(), NCV_ALLOCATOR_BAD_ALLOC);
NCVMatrixAlloc<Ncv8u> h_src(cpuAllocator, width, height);
ncvAssertReturn(h_src.isMemAllocated(), NCV_ALLOCATOR_BAD_ALLOC);
NCVVectorAlloc<NcvRect32u> d_rects(gpuAllocator, 100);
ncvAssertReturn(d_rects.isMemAllocated(), NCV_ALLOCATOR_BAD_ALLOC);
NCV_SKIP_COND_BEGIN
for (Ncv32u i=0; i<(Ncv32u)srcdst->rows; i++)
{
memcpy(h_src.ptr() + i * h_src.stride(), srcdst->ptr(i), srcdst->cols);
}
ncvStat = h_src.copySolid(d_src, 0);
ncvAssertReturnNcvStat(ncvStat);
ncvAssertCUDAReturn(cudaStreamSynchronize(0), NCV_CUDA_ERROR);
NCV_SKIP_COND_END
NcvSize32u roi;
roi.width = d_src.width();
roi.height = d_src.height();
Ncv32u numDetections;
ncvStat = ncvDetectObjectsMultiScale_device(
d_src, roi, d_rects, numDetections, haar, h_haarStages,
d_haarStages, d_haarNodes, d_haarFeatures,
haar.ClassifierSize,
(bFilterRects || bLargestFace) ? 4 : 0,
1.2f, 1,
(bLargestFace ? NCVPipeObjDet_FindLargestObject : 0)
| NCVPipeObjDet_VisualizeInPlace,
gpuAllocator, cpuAllocator, devProp, 0);
ncvAssertReturnNcvStat(ncvStat);
ncvAssertCUDAReturn(cudaStreamSynchronize(0), NCV_CUDA_ERROR);
NCV_SKIP_COND_BEGIN
ncvStat = d_src.copySolid(h_src, 0);
ncvAssertReturnNcvStat(ncvStat);
ncvAssertCUDAReturn(cudaStreamSynchronize(0), NCV_CUDA_ERROR);
for (Ncv32u i=0; i<(Ncv32u)srcdst->rows; i++)
{
memcpy(srcdst->ptr(i), h_src.ptr() + i * h_src.stride(), srcdst->cols);
}
NCV_SKIP_COND_END
return NCV_SUCCESS;
}
int main(int argc, const char** argv)
{
cout << "OpenCV / NVIDIA Computer Vision" << endl;
cout << "Face Detection in video and live feed" << endl;
cout << "Syntax: exename <cascade_file> <image_or_video_or_cameraid>" << endl;
cout << "=========================================" << endl;
ncvAssertPrintReturn(cv::gpu::getCudaEnabledDeviceCount() != 0, "No GPU found or the library is compiled without GPU support", -1);
ncvAssertPrintReturn(argc == 3, "Invalid number of arguments", -1);
cv::gpu::printShortCudaDeviceInfo(cv::gpu::getDevice());
string cascadeName = argv[1];
string inputName = argv[2];
NCVStatus ncvStat;
NcvBool bQuit = false;
VideoCapture capture;
Size2i frameSize;
//open content source
Mat image = imread(inputName);
Mat frame;
if (!image.empty())
{
frameSize.width = image.cols;
frameSize.height = image.rows;
}
else
{
if (!capture.open(inputName))
{
int camid = -1;
istringstream ss(inputName);
int x = 0;
ss >> x;
ncvAssertPrintReturn(capture.open(camid) != 0, "Can't open source", -1);
}
capture >> frame;
ncvAssertPrintReturn(!frame.empty(), "Empty video source", -1);
frameSize.width = frame.cols;
frameSize.height = frame.rows;
}
NcvBool bUseGPU = true;
NcvBool bLargestObject = false;
NcvBool bFilterRects = true;
NcvBool bHelpScreen = false;
CascadeClassifier classifierOpenCV;
ncvAssertPrintReturn(classifierOpenCV.load(cascadeName) != 0, "Error (in OpenCV) opening classifier", -1);
int devId;
ncvAssertCUDAReturn(cudaGetDevice(&devId), -1);
cudaDeviceProp devProp;
ncvAssertCUDAReturn(cudaGetDeviceProperties(&devProp, devId), -1);
cout << "Using GPU: " << devId << "(" << devProp.name <<
"), arch=" << devProp.major << "." << devProp.minor << endl;
//==============================================================================
//
// Load the classifier from file (assuming its size is about 1 mb)
// using a simple allocator
//
//==============================================================================
NCVMemNativeAllocator gpuCascadeAllocator(NCVMemoryTypeDevice, static_cast<Ncv32u>(devProp.textureAlignment));
ncvAssertPrintReturn(gpuCascadeAllocator.isInitialized(), "Error creating cascade GPU allocator", -1);
NCVMemNativeAllocator cpuCascadeAllocator(NCVMemoryTypeHostPinned, static_cast<Ncv32u>(devProp.textureAlignment));
ncvAssertPrintReturn(cpuCascadeAllocator.isInitialized(), "Error creating cascade CPU allocator", -1);
Ncv32u haarNumStages, haarNumNodes, haarNumFeatures;
ncvStat = ncvHaarGetClassifierSize(cascadeName, haarNumStages, haarNumNodes, haarNumFeatures);
ncvAssertPrintReturn(ncvStat == NCV_SUCCESS, "Error reading classifier size (check the file)", -1);
NCVVectorAlloc<HaarStage64> h_haarStages(cpuCascadeAllocator, haarNumStages);
ncvAssertPrintReturn(h_haarStages.isMemAllocated(), "Error in cascade CPU allocator", -1);
NCVVectorAlloc<HaarClassifierNode128> h_haarNodes(cpuCascadeAllocator, haarNumNodes);
ncvAssertPrintReturn(h_haarNodes.isMemAllocated(), "Error in cascade CPU allocator", -1);
NCVVectorAlloc<HaarFeature64> h_haarFeatures(cpuCascadeAllocator, haarNumFeatures);
ncvAssertPrintReturn(h_haarFeatures.isMemAllocated(), "Error in cascade CPU allocator", -1);
HaarClassifierCascadeDescriptor haar;
ncvStat = ncvHaarLoadFromFile_host(cascadeName, haar, h_haarStages, h_haarNodes, h_haarFeatures);
ncvAssertPrintReturn(ncvStat == NCV_SUCCESS, "Error loading classifier", -1);
NCVVectorAlloc<HaarStage64> d_haarStages(gpuCascadeAllocator, haarNumStages);
ncvAssertPrintReturn(d_haarStages.isMemAllocated(), "Error in cascade GPU allocator", -1);
NCVVectorAlloc<HaarClassifierNode128> d_haarNodes(gpuCascadeAllocator, haarNumNodes);
ncvAssertPrintReturn(d_haarNodes.isMemAllocated(), "Error in cascade GPU allocator", -1);
NCVVectorAlloc<HaarFeature64> d_haarFeatures(gpuCascadeAllocator, haarNumFeatures);
ncvAssertPrintReturn(d_haarFeatures.isMemAllocated(), "Error in cascade GPU allocator", -1);
ncvStat = h_haarStages.copySolid(d_haarStages, 0);
ncvAssertPrintReturn(ncvStat == NCV_SUCCESS, "Error copying cascade to GPU", -1);
ncvStat = h_haarNodes.copySolid(d_haarNodes, 0);
ncvAssertPrintReturn(ncvStat == NCV_SUCCESS, "Error copying cascade to GPU", -1);
ncvStat = h_haarFeatures.copySolid(d_haarFeatures, 0);
ncvAssertPrintReturn(ncvStat == NCV_SUCCESS, "Error copying cascade to GPU", -1);
//==============================================================================
//
// Calculate memory requirements and create real allocators
//
//==============================================================================
NCVMemStackAllocator gpuCounter(static_cast<Ncv32u>(devProp.textureAlignment));
ncvAssertPrintReturn(gpuCounter.isInitialized(), "Error creating GPU memory counter", -1);
NCVMemStackAllocator cpuCounter(static_cast<Ncv32u>(devProp.textureAlignment));
ncvAssertPrintReturn(cpuCounter.isInitialized(), "Error creating CPU memory counter", -1);
ncvStat = process(NULL, frameSize.width, frameSize.height,
false, false, haar,
d_haarStages, d_haarNodes,
d_haarFeatures, h_haarStages,
gpuCounter, cpuCounter, devProp);
ncvAssertPrintReturn(ncvStat == NCV_SUCCESS, "Error in memory counting pass", -1);
NCVMemStackAllocator gpuAllocator(NCVMemoryTypeDevice, gpuCounter.maxSize(), static_cast<Ncv32u>(devProp.textureAlignment));
ncvAssertPrintReturn(gpuAllocator.isInitialized(), "Error creating GPU memory allocator", -1);
NCVMemStackAllocator cpuAllocator(NCVMemoryTypeHostPinned, cpuCounter.maxSize(), static_cast<Ncv32u>(devProp.textureAlignment));
ncvAssertPrintReturn(cpuAllocator.isInitialized(), "Error creating CPU memory allocator", -1);
printf("Initialized for frame size [%dx%d]\n", frameSize.width, frameSize.height);
//==============================================================================
//
// Main processing loop
//
//==============================================================================
namedWindow(wndTitle, 1);
Mat gray, frameDisp;
do
{
Mat gray;
cvtColor((image.empty() ? frame : image), gray, CV_BGR2GRAY);
//
// process
//
NcvSize32u minSize = haar.ClassifierSize;
if (bLargestObject)
{
Ncv32u ratioX = preferredVideoFrameSize.width / minSize.width;
Ncv32u ratioY = preferredVideoFrameSize.height / minSize.height;
Ncv32u ratioSmallest = min(ratioX, ratioY);
ratioSmallest = max((Ncv32u)(ratioSmallest / 2.5f), (Ncv32u)1);
minSize.width *= ratioSmallest;
minSize.height *= ratioSmallest;
}
Ncv32f avgTime;
NcvTimer timer = ncvStartTimer();
if (bUseGPU)
{
ncvStat = process(&gray, frameSize.width, frameSize.height,
bFilterRects, bLargestObject, haar,
d_haarStages, d_haarNodes,
d_haarFeatures, h_haarStages,
gpuAllocator, cpuAllocator, devProp);
ncvAssertPrintReturn(ncvStat == NCV_SUCCESS, "Error in memory counting pass", -1);
}
else
{
vector<Rect> rectsOpenCV;
classifierOpenCV.detectMultiScale(
gray,
rectsOpenCV,
1.2f,
bFilterRects ? 4 : 0,
(bLargestObject ? CV_HAAR_FIND_BIGGEST_OBJECT : 0)
| CV_HAAR_SCALE_IMAGE,
Size(minSize.width, minSize.height));
for (size_t rt = 0; rt < rectsOpenCV.size(); ++rt)
rectangle(gray, rectsOpenCV[rt], Scalar(255));
}
avgTime = (Ncv32f)ncvEndQueryTimerMs(timer);
cvtColor(gray, frameDisp, CV_GRAY2BGR);
displayState(frameDisp, bHelpScreen, bUseGPU, bLargestObject, bFilterRects, 1000.0f / avgTime);
imshow(wndTitle, frameDisp);
//handle input
switch (cvWaitKey(3))
{
case ' ':
bUseGPU = !bUseGPU;
break;
case 'm':
case 'M':
bLargestObject = !bLargestObject;
break;
case 'f':
case 'F':
bFilterRects = !bFilterRects;
break;
case 'h':
case 'H':
bHelpScreen = !bHelpScreen;
break;
case 27:
bQuit = true;
break;
}
// For camera and video file, capture the next image
if (capture.isOpened())
{
capture >> frame;
if (frame.empty())
{
break;
}
}
} while (!bQuit);
cvDestroyWindow(wndTitle.c_str());
return 0;
}
#endif //!defined(HAVE_CUDA)
#if defined _MSC_VER && _MSC_VER >= 1400
#pragma warning( disable : 4201 4408 4127 4100)
#endif
#include "cvconfig.h"
#include <iostream>
#include <iomanip>
#include <cstdio>
#include "opencv2/gpu/gpu.hpp"
#include "opencv2/highgui/highgui.hpp"
#ifdef HAVE_CUDA
#include "NCVHaarObjectDetection.hpp"
#endif
using namespace std;
using namespace cv;
#if !defined(HAVE_CUDA)
int main( int, const char** )
{
cout << "Please compile the library with CUDA support" << endl;
return -1;
}
#else
const Size2i preferredVideoFrameSize(640, 480);
const string wndTitle = "NVIDIA Computer Vision :: Haar Classifiers Cascade";
void matPrint(Mat &img, int lineOffsY, Scalar fontColor, const string &ss)
{
int fontFace = FONT_HERSHEY_DUPLEX;
double fontScale = 0.8;
int fontThickness = 2;
Size fontSize = cv::getTextSize("T[]", fontFace, fontScale, fontThickness, 0);
Point org;
org.x = 1;
org.y = 3 * fontSize.height * (lineOffsY + 1) / 2;
putText(img, ss, org, fontFace, fontScale, CV_RGB(0,0,0), 5*fontThickness/2, 16);
putText(img, ss, org, fontFace, fontScale, fontColor, fontThickness, 16);
}
void displayState(Mat &canvas, bool bHelp, bool bGpu, bool bLargestFace, bool bFilter, double fps)
{
Scalar fontColorRed = CV_RGB(255,0,0);
Scalar fontColorNV = CV_RGB(118,185,0);
ostringstream ss;
ss << "FPS = " << setprecision(1) << fixed << fps;
matPrint(canvas, 0, fontColorRed, ss.str());
ss.str("");
ss << "[" << canvas.cols << "x" << canvas.rows << "], " <<
(bGpu ? "GPU, " : "CPU, ") <<
(bLargestFace ? "OneFace, " : "MultiFace, ") <<
(bFilter ? "Filter:ON" : "Filter:OFF");
matPrint(canvas, 1, fontColorRed, ss.str());
if (bHelp)
{
matPrint(canvas, 2, fontColorNV, "Space - switch GPU / CPU");
matPrint(canvas, 3, fontColorNV, "M - switch OneFace / MultiFace");
matPrint(canvas, 4, fontColorNV, "F - toggle rectangles Filter");
matPrint(canvas, 5, fontColorNV, "H - toggle hotkeys help");
}
else
{
matPrint(canvas, 2, fontColorNV, "H - toggle hotkeys help");
}
}
NCVStatus process(Mat *srcdst,
Ncv32u width, Ncv32u height,
NcvBool bFilterRects, NcvBool bLargestFace,
HaarClassifierCascadeDescriptor &haar,
NCVVector<HaarStage64> &d_haarStages, NCVVector<HaarClassifierNode128> &d_haarNodes,
NCVVector<HaarFeature64> &d_haarFeatures, NCVVector<HaarStage64> &h_haarStages,
INCVMemAllocator &gpuAllocator,
INCVMemAllocator &cpuAllocator,
cudaDeviceProp &devProp)
{
ncvAssertReturn(!((srcdst == NULL) ^ gpuAllocator.isCounting()), NCV_NULL_PTR);
NCVStatus ncvStat;
NCV_SET_SKIP_COND(gpuAllocator.isCounting());
NCVMatrixAlloc<Ncv8u> d_src(gpuAllocator, width, height);
ncvAssertReturn(d_src.isMemAllocated(), NCV_ALLOCATOR_BAD_ALLOC);
NCVMatrixAlloc<Ncv8u> h_src(cpuAllocator, width, height);
ncvAssertReturn(h_src.isMemAllocated(), NCV_ALLOCATOR_BAD_ALLOC);
NCVVectorAlloc<NcvRect32u> d_rects(gpuAllocator, 100);
ncvAssertReturn(d_rects.isMemAllocated(), NCV_ALLOCATOR_BAD_ALLOC);
NCV_SKIP_COND_BEGIN
for (Ncv32u i=0; i<(Ncv32u)srcdst->rows; i++)
{
memcpy(h_src.ptr() + i * h_src.stride(), srcdst->ptr(i), srcdst->cols);
}
ncvStat = h_src.copySolid(d_src, 0);
ncvAssertReturnNcvStat(ncvStat);
ncvAssertCUDAReturn(cudaStreamSynchronize(0), NCV_CUDA_ERROR);
NCV_SKIP_COND_END
NcvSize32u roi;
roi.width = d_src.width();
roi.height = d_src.height();
Ncv32u numDetections;
ncvStat = ncvDetectObjectsMultiScale_device(
d_src, roi, d_rects, numDetections, haar, h_haarStages,
d_haarStages, d_haarNodes, d_haarFeatures,
haar.ClassifierSize,
(bFilterRects || bLargestFace) ? 4 : 0,
1.2f, 1,
(bLargestFace ? NCVPipeObjDet_FindLargestObject : 0)
| NCVPipeObjDet_VisualizeInPlace,
gpuAllocator, cpuAllocator, devProp, 0);
ncvAssertReturnNcvStat(ncvStat);
ncvAssertCUDAReturn(cudaStreamSynchronize(0), NCV_CUDA_ERROR);
NCV_SKIP_COND_BEGIN
ncvStat = d_src.copySolid(h_src, 0);
ncvAssertReturnNcvStat(ncvStat);
ncvAssertCUDAReturn(cudaStreamSynchronize(0), NCV_CUDA_ERROR);
for (Ncv32u i=0; i<(Ncv32u)srcdst->rows; i++)
{
memcpy(srcdst->ptr(i), h_src.ptr() + i * h_src.stride(), srcdst->cols);
}
NCV_SKIP_COND_END
return NCV_SUCCESS;
}
int main(int argc, const char** argv)
{
cout << "OpenCV / NVIDIA Computer Vision" << endl;
cout << "Face Detection in video and live feed" << endl;
cout << "Syntax: exename <cascade_file> <image_or_video_or_cameraid>" << endl;
cout << "=========================================" << endl;
ncvAssertPrintReturn(cv::gpu::getCudaEnabledDeviceCount() != 0, "No GPU found or the library is compiled without GPU support", -1);
ncvAssertPrintReturn(argc == 3, "Invalid number of arguments", -1);
cv::gpu::printShortCudaDeviceInfo(cv::gpu::getDevice());
string cascadeName = argv[1];
string inputName = argv[2];
NCVStatus ncvStat;
NcvBool bQuit = false;
VideoCapture capture;
Size2i frameSize;
//open content source
Mat image = imread(inputName);
Mat frame;
if (!image.empty())
{
frameSize.width = image.cols;
frameSize.height = image.rows;
}
else
{
if (!capture.open(inputName))
{
int camid = -1;
istringstream ss(inputName);
int x = 0;
ss >> x;
ncvAssertPrintReturn(capture.open(camid) != 0, "Can't open source", -1);
}
capture >> frame;
ncvAssertPrintReturn(!frame.empty(), "Empty video source", -1);
frameSize.width = frame.cols;
frameSize.height = frame.rows;
}
NcvBool bUseGPU = true;
NcvBool bLargestObject = false;
NcvBool bFilterRects = true;
NcvBool bHelpScreen = false;
CascadeClassifier classifierOpenCV;
ncvAssertPrintReturn(classifierOpenCV.load(cascadeName) != 0, "Error (in OpenCV) opening classifier", -1);
int devId;
ncvAssertCUDAReturn(cudaGetDevice(&devId), -1);
cudaDeviceProp devProp;
ncvAssertCUDAReturn(cudaGetDeviceProperties(&devProp, devId), -1);
cout << "Using GPU: " << devId << "(" << devProp.name <<
"), arch=" << devProp.major << "." << devProp.minor << endl;
//==============================================================================
//
// Load the classifier from file (assuming its size is about 1 mb)
// using a simple allocator
//
//==============================================================================
NCVMemNativeAllocator gpuCascadeAllocator(NCVMemoryTypeDevice, static_cast<Ncv32u>(devProp.textureAlignment));
ncvAssertPrintReturn(gpuCascadeAllocator.isInitialized(), "Error creating cascade GPU allocator", -1);
NCVMemNativeAllocator cpuCascadeAllocator(NCVMemoryTypeHostPinned, static_cast<Ncv32u>(devProp.textureAlignment));
ncvAssertPrintReturn(cpuCascadeAllocator.isInitialized(), "Error creating cascade CPU allocator", -1);
Ncv32u haarNumStages, haarNumNodes, haarNumFeatures;
ncvStat = ncvHaarGetClassifierSize(cascadeName, haarNumStages, haarNumNodes, haarNumFeatures);
ncvAssertPrintReturn(ncvStat == NCV_SUCCESS, "Error reading classifier size (check the file)", -1);
NCVVectorAlloc<HaarStage64> h_haarStages(cpuCascadeAllocator, haarNumStages);
ncvAssertPrintReturn(h_haarStages.isMemAllocated(), "Error in cascade CPU allocator", -1);
NCVVectorAlloc<HaarClassifierNode128> h_haarNodes(cpuCascadeAllocator, haarNumNodes);
ncvAssertPrintReturn(h_haarNodes.isMemAllocated(), "Error in cascade CPU allocator", -1);
NCVVectorAlloc<HaarFeature64> h_haarFeatures(cpuCascadeAllocator, haarNumFeatures);
ncvAssertPrintReturn(h_haarFeatures.isMemAllocated(), "Error in cascade CPU allocator", -1);
HaarClassifierCascadeDescriptor haar;
ncvStat = ncvHaarLoadFromFile_host(cascadeName, haar, h_haarStages, h_haarNodes, h_haarFeatures);
ncvAssertPrintReturn(ncvStat == NCV_SUCCESS, "Error loading classifier", -1);
NCVVectorAlloc<HaarStage64> d_haarStages(gpuCascadeAllocator, haarNumStages);
ncvAssertPrintReturn(d_haarStages.isMemAllocated(), "Error in cascade GPU allocator", -1);
NCVVectorAlloc<HaarClassifierNode128> d_haarNodes(gpuCascadeAllocator, haarNumNodes);
ncvAssertPrintReturn(d_haarNodes.isMemAllocated(), "Error in cascade GPU allocator", -1);
NCVVectorAlloc<HaarFeature64> d_haarFeatures(gpuCascadeAllocator, haarNumFeatures);
ncvAssertPrintReturn(d_haarFeatures.isMemAllocated(), "Error in cascade GPU allocator", -1);
ncvStat = h_haarStages.copySolid(d_haarStages, 0);
ncvAssertPrintReturn(ncvStat == NCV_SUCCESS, "Error copying cascade to GPU", -1);
ncvStat = h_haarNodes.copySolid(d_haarNodes, 0);
ncvAssertPrintReturn(ncvStat == NCV_SUCCESS, "Error copying cascade to GPU", -1);
ncvStat = h_haarFeatures.copySolid(d_haarFeatures, 0);
ncvAssertPrintReturn(ncvStat == NCV_SUCCESS, "Error copying cascade to GPU", -1);
//==============================================================================
//
// Calculate memory requirements and create real allocators
//
//==============================================================================
NCVMemStackAllocator gpuCounter(static_cast<Ncv32u>(devProp.textureAlignment));
ncvAssertPrintReturn(gpuCounter.isInitialized(), "Error creating GPU memory counter", -1);
NCVMemStackAllocator cpuCounter(static_cast<Ncv32u>(devProp.textureAlignment));
ncvAssertPrintReturn(cpuCounter.isInitialized(), "Error creating CPU memory counter", -1);
ncvStat = process(NULL, frameSize.width, frameSize.height,
false, false, haar,
d_haarStages, d_haarNodes,
d_haarFeatures, h_haarStages,
gpuCounter, cpuCounter, devProp);
ncvAssertPrintReturn(ncvStat == NCV_SUCCESS, "Error in memory counting pass", -1);
NCVMemStackAllocator gpuAllocator(NCVMemoryTypeDevice, gpuCounter.maxSize(), static_cast<Ncv32u>(devProp.textureAlignment));
ncvAssertPrintReturn(gpuAllocator.isInitialized(), "Error creating GPU memory allocator", -1);
NCVMemStackAllocator cpuAllocator(NCVMemoryTypeHostPinned, cpuCounter.maxSize(), static_cast<Ncv32u>(devProp.textureAlignment));
ncvAssertPrintReturn(cpuAllocator.isInitialized(), "Error creating CPU memory allocator", -1);
printf("Initialized for frame size [%dx%d]\n", frameSize.width, frameSize.height);
//==============================================================================
//
// Main processing loop
//
//==============================================================================
namedWindow(wndTitle, 1);
Mat gray, frameDisp;
do
{
Mat gray;
cvtColor((image.empty() ? frame : image), gray, CV_BGR2GRAY);
//
// process
//
NcvSize32u minSize = haar.ClassifierSize;
if (bLargestObject)
{
Ncv32u ratioX = preferredVideoFrameSize.width / minSize.width;
Ncv32u ratioY = preferredVideoFrameSize.height / minSize.height;
Ncv32u ratioSmallest = min(ratioX, ratioY);
ratioSmallest = max((Ncv32u)(ratioSmallest / 2.5f), (Ncv32u)1);
minSize.width *= ratioSmallest;
minSize.height *= ratioSmallest;
}
Ncv32f avgTime;
NcvTimer timer = ncvStartTimer();
if (bUseGPU)
{
ncvStat = process(&gray, frameSize.width, frameSize.height,
bFilterRects, bLargestObject, haar,
d_haarStages, d_haarNodes,
d_haarFeatures, h_haarStages,
gpuAllocator, cpuAllocator, devProp);
ncvAssertPrintReturn(ncvStat == NCV_SUCCESS, "Error in memory counting pass", -1);
}
else
{
vector<Rect> rectsOpenCV;
classifierOpenCV.detectMultiScale(
gray,
rectsOpenCV,
1.2f,
bFilterRects ? 4 : 0,
(bLargestObject ? CV_HAAR_FIND_BIGGEST_OBJECT : 0)
| CV_HAAR_SCALE_IMAGE,
Size(minSize.width, minSize.height));
for (size_t rt = 0; rt < rectsOpenCV.size(); ++rt)
rectangle(gray, rectsOpenCV[rt], Scalar(255));
}
avgTime = (Ncv32f)ncvEndQueryTimerMs(timer);
cvtColor(gray, frameDisp, CV_GRAY2BGR);
displayState(frameDisp, bHelpScreen, bUseGPU, bLargestObject, bFilterRects, 1000.0f / avgTime);
imshow(wndTitle, frameDisp);
//handle input
switch (cvWaitKey(3))
{
case ' ':
bUseGPU = !bUseGPU;
break;
case 'm':
case 'M':
bLargestObject = !bLargestObject;
break;
case 'f':
case 'F':
bFilterRects = !bFilterRects;
break;
case 'h':
case 'H':
bHelpScreen = !bHelpScreen;
break;
case 27:
bQuit = true;
break;
}
// For camera and video file, capture the next image
if (capture.isOpened())
{
capture >> frame;
if (frame.empty())
{
break;
}
}
} while (!bQuit);
cvDestroyWindow(wndTitle.c_str());
return 0;
}
#endif //!defined(HAVE_CUDA)

304
samples/gpu/driver_api_multi.cpp
View File

@@ -1,152 +1,152 @@
/* This sample demonstrates the way you can perform independed tasks
on the different GPUs */
// Disable some warnings which are caused with CUDA headers
#if defined(_MSC_VER)
#pragma warning(disable: 4201 4408 4100)
#endif
#include <iostream>
#include "cvconfig.h"
#include "opencv2/core/core.hpp"
#include "opencv2/gpu/gpu.hpp"
#if !defined(HAVE_CUDA) || !defined(HAVE_TBB)
int main()
{
#if !defined(HAVE_CUDA)
std::cout << "CUDA support is required (CMake key 'WITH_CUDA' must be true).\n";
#endif
#if !defined(HAVE_TBB)
std::cout << "TBB support is required (CMake key 'WITH_TBB' must be true).\n";
#endif
return 0;
}
#else
#include <cuda.h>
#include <cuda_runtime.h>
#include "opencv2/core/internal.hpp" // For TBB wrappers
using namespace std;
using namespace cv;
using namespace cv::gpu;
struct Worker { void operator()(int device_id) const; };
void destroyContexts();
#define safeCall(expr) safeCall_(expr, #expr, __FILE__, __LINE__)
inline void safeCall_(int code, const char* expr, const char* file, int line)
{
if (code != CUDA_SUCCESS)
{
std::cout << "CUDA driver API error: code " << code << ", expr " << expr
<< ", file " << file << ", line " << line << endl;
destroyContexts();
exit(-1);
}
}
// Each GPU is associated with its own context
CUcontext contexts[2];
int main(int argc, char **argv)
{
if (argc > 1)
{
cout << "CUDA driver API sample\n";
return -1;
}
int num_devices = getCudaEnabledDeviceCount();
if (num_devices < 2)
{
std::cout << "Two or more GPUs are required\n";
return -1;
}
for (int i = 0; i < num_devices; ++i)
{
cv::gpu::printShortCudaDeviceInfo(i);
DeviceInfo dev_info(i);
if (!dev_info.isCompatible())
{
std::cout << "GPU module isn't built for GPU #" << i << " ("
<< dev_info.name() << ", CC " << dev_info.majorVersion()
<< dev_info.minorVersion() << "\n";
return -1;
}
}
// Init CUDA Driver API
safeCall(cuInit(0));
// Create context for GPU #0
CUdevice device;
safeCall(cuDeviceGet(&device, 0));
safeCall(cuCtxCreate(&contexts[0], 0, device));
CUcontext prev_context;
safeCall(cuCtxPopCurrent(&prev_context));
// Create context for GPU #1
safeCall(cuDeviceGet(&device, 1));
safeCall(cuCtxCreate(&contexts[1], 0, device));
safeCall(cuCtxPopCurrent(&prev_context));
// Execute calculation in two threads using two GPUs
int devices[] = {0, 1};
parallel_do(devices, devices + 2, Worker());
destroyContexts();
return 0;
}
void Worker::operator()(int device_id) const
{
// Set the proper context
safeCall(cuCtxPushCurrent(contexts[device_id]));
Mat src(1000, 1000, CV_32F);
Mat dst;
RNG rng(0);
rng.fill(src, RNG::UNIFORM, 0, 1);
// CPU works
transpose(src, dst);
// GPU works
GpuMat d_src(src);
GpuMat d_dst;
transpose(d_src, d_dst);
// Check results
bool passed = norm(dst - Mat(d_dst), NORM_INF) < 1e-3;
std::cout << "GPU #" << device_id << " (" << DeviceInfo().name() << "): "
<< (passed ? "passed" : "FAILED") << endl;
// Deallocate data here, otherwise deallocation will be performed
// after context is extracted from the stack
d_src.release();
d_dst.release();
CUcontext prev_context;
safeCall(cuCtxPopCurrent(&prev_context));
}
void destroyContexts()
{
safeCall(cuCtxDestroy(contexts[0]));
safeCall(cuCtxDestroy(contexts[1]));
}
#endif
/* This sample demonstrates the way you can perform independed tasks
on the different GPUs */
// Disable some warnings which are caused with CUDA headers
#if defined(_MSC_VER)
#pragma warning(disable: 4201 4408 4100)
#endif
#include <iostream>
#include "cvconfig.h"
#include "opencv2/core/core.hpp"
#include "opencv2/gpu/gpu.hpp"
#if !defined(HAVE_CUDA) || !defined(HAVE_TBB)
int main()
{
#if !defined(HAVE_CUDA)
std::cout << "CUDA support is required (CMake key 'WITH_CUDA' must be true).\n";
#endif
#if !defined(HAVE_TBB)
std::cout << "TBB support is required (CMake key 'WITH_TBB' must be true).\n";
#endif
return 0;
}
#else
#include <cuda.h>
#include <cuda_runtime.h>
#include "opencv2/core/internal.hpp" // For TBB wrappers
using namespace std;
using namespace cv;
using namespace cv::gpu;
struct Worker { void operator()(int device_id) const; };
void destroyContexts();
#define safeCall(expr) safeCall_(expr, #expr, __FILE__, __LINE__)
inline void safeCall_(int code, const char* expr, const char* file, int line)
{
if (code != CUDA_SUCCESS)
{
std::cout << "CUDA driver API error: code " << code << ", expr " << expr
<< ", file " << file << ", line " << line << endl;
destroyContexts();
exit(-1);
}
}
// Each GPU is associated with its own context
CUcontext contexts[2];
int main(int argc, char **argv)
{
if (argc > 1)
{
cout << "CUDA driver API sample\n";
return -1;
}
int num_devices = getCudaEnabledDeviceCount();
if (num_devices < 2)
{
std::cout << "Two or more GPUs are required\n";
return -1;
}
for (int i = 0; i < num_devices; ++i)
{
cv::gpu::printShortCudaDeviceInfo(i);
DeviceInfo dev_info(i);
if (!dev_info.isCompatible())
{
std::cout << "GPU module isn't built for GPU #" << i << " ("
<< dev_info.name() << ", CC " << dev_info.majorVersion()
<< dev_info.minorVersion() << "\n";
return -1;
}
}
// Init CUDA Driver API
safeCall(cuInit(0));
// Create context for GPU #0
CUdevice device;
safeCall(cuDeviceGet(&device, 0));
safeCall(cuCtxCreate(&contexts[0], 0, device));
CUcontext prev_context;
safeCall(cuCtxPopCurrent(&prev_context));
// Create context for GPU #1
safeCall(cuDeviceGet(&device, 1));
safeCall(cuCtxCreate(&contexts[1], 0, device));
safeCall(cuCtxPopCurrent(&prev_context));
// Execute calculation in two threads using two GPUs
int devices[] = {0, 1};
parallel_do(devices, devices + 2, Worker());
destroyContexts();
return 0;
}
void Worker::operator()(int device_id) const
{
// Set the proper context
safeCall(cuCtxPushCurrent(contexts[device_id]));
Mat src(1000, 1000, CV_32F);
Mat dst;
RNG rng(0);
rng.fill(src, RNG::UNIFORM, 0, 1);
// CPU works
transpose(src, dst);
// GPU works
GpuMat d_src(src);
GpuMat d_dst;
transpose(d_src, d_dst);
// Check results
bool passed = norm(dst - Mat(d_dst), NORM_INF) < 1e-3;
std::cout << "GPU #" << device_id << " (" << DeviceInfo().name() << "): "
<< (passed ? "passed" : "FAILED") << endl;
// Deallocate data here, otherwise deallocation will be performed
// after context is extracted from the stack
d_src.release();
d_dst.release();
CUcontext prev_context;
safeCall(cuCtxPopCurrent(&prev_context));
}
void destroyContexts()
{
safeCall(cuCtxDestroy(contexts[0]));
safeCall(cuCtxDestroy(contexts[1]));
}
#endif

422
samples/gpu/driver_api_stereo_multi.cpp
View File

@@ -1,211 +1,211 @@
/* This sample demonstrates working on one piece of data using two GPUs.
It splits input into two parts and processes them separately on different
GPUs. */
// Disable some warnings which are caused with CUDA headers
#if defined(_MSC_VER)
#pragma warning(disable: 4201 4408 4100)
#endif
#include <iostream>
#include "cvconfig.h"
#include "opencv2/core/core.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/gpu/gpu.hpp"
#if !defined(HAVE_CUDA) || !defined(HAVE_TBB)
int main()
{
#if !defined(HAVE_CUDA)
std::cout << "CUDA support is required (CMake key 'WITH_CUDA' must be true).\n";
#endif
#if !defined(HAVE_TBB)
std::cout << "TBB support is required (CMake key 'WITH_TBB' must be true).\n";
#endif
return 0;
}
#else
#include <cuda.h>
#include <cuda_runtime.h>
#include "opencv2/core/internal.hpp" // For TBB wrappers
using namespace std;
using namespace cv;
using namespace cv::gpu;
struct Worker { void operator()(int device_id) const; };
void destroyContexts();
#define safeCall(expr) safeCall_(expr, #expr, __FILE__, __LINE__)
inline void safeCall_(int code, const char* expr, const char* file, int line)
{
if (code != CUDA_SUCCESS)
{
std::cout << "CUDA driver API error: code " << code << ", expr " << expr
<< ", file " << file << ", line " << line << endl;
destroyContexts();
exit(-1);
}
}
// Each GPU is associated with its own context
CUcontext contexts[2];
void inline contextOn(int id)
{
safeCall(cuCtxPushCurrent(contexts[id]));
}
void inline contextOff()
{
CUcontext prev_context;
safeCall(cuCtxPopCurrent(&prev_context));
}
// GPUs data
GpuMat d_left[2];
GpuMat d_right[2];
StereoBM_GPU* bm[2];
GpuMat d_result[2];
// CPU result
Mat result;
void printHelp()
{
std::cout << "Usage: driver_api_stereo_multi_gpu --left <left_image> --right <right_image>\n";
}
int main(int argc, char** argv)
{
if (argc < 5)
{
printHelp();
return -1;
}
int num_devices = getCudaEnabledDeviceCount();
if (num_devices < 2)
{
std::cout << "Two or more GPUs are required\n";
return -1;
}
for (int i = 0; i < num_devices; ++i)
{
cv::gpu::printShortCudaDeviceInfo(i);
DeviceInfo dev_info(i);
if (!dev_info.isCompatible())
{
std::cout << "GPU module isn't built for GPU #" << i << " ("
<< dev_info.name() << ", CC " << dev_info.majorVersion()
<< dev_info.minorVersion() << "\n";
return -1;
}
}
// Load input data
Mat left, right;
for (int i = 1; i < argc; ++i)
{
if (string(argv[i]) == "--left")
{
left = imread(argv[++i], CV_LOAD_IMAGE_GRAYSCALE);
CV_Assert(!left.empty());
}
else if (string(argv[i]) == "--right")
{
right = imread(argv[++i], CV_LOAD_IMAGE_GRAYSCALE);
CV_Assert(!right.empty());
}
else if (string(argv[i]) == "--help")
{
printHelp();
return -1;
}
}
// Init CUDA Driver API
safeCall(cuInit(0));
// Create context for GPU #0
CUdevice device;
safeCall(cuDeviceGet(&device, 0));
safeCall(cuCtxCreate(&contexts[0], 0, device));
contextOff();
// Create context for GPU #1
safeCall(cuDeviceGet(&device, 1));
safeCall(cuCtxCreate(&contexts[1], 0, device));
contextOff();
// Split source images for processing on GPU #0
contextOn(0);
d_left[0].upload(left.rowRange(0, left.rows / 2));
d_right[0].upload(right.rowRange(0, right.rows / 2));
bm[0] = new StereoBM_GPU();
contextOff();
// Split source images for processing on the GPU #1
contextOn(1);
d_left[1].upload(left.rowRange(left.rows / 2, left.rows));
d_right[1].upload(right.rowRange(right.rows / 2, right.rows));
bm[1] = new StereoBM_GPU();
contextOff();
// Execute calculation in two threads using two GPUs
int devices[] = {0, 1};
parallel_do(devices, devices + 2, Worker());
// Release the first GPU resources
contextOn(0);
imshow("GPU #0 result", Mat(d_result[0]));
d_left[0].release();
d_right[0].release();
d_result[0].release();
delete bm[0];
contextOff();
// Release the second GPU resources
contextOn(1);
imshow("GPU #1 result", Mat(d_result[1]));
d_left[1].release();
d_right[1].release();
d_result[1].release();
delete bm[1];
contextOff();
waitKey();
destroyContexts();
return 0;
}
void Worker::operator()(int device_id) const
{
contextOn(device_id);
bm[device_id]->operator()(d_left[device_id], d_right[device_id],
d_result[device_id]);
std::cout << "GPU #" << device_id << " (" << DeviceInfo().name()
<< "): finished\n";
contextOff();
}
void destroyContexts()
{
safeCall(cuCtxDestroy(contexts[0]));
safeCall(cuCtxDestroy(contexts[1]));
}
#endif
/* This sample demonstrates working on one piece of data using two GPUs.
It splits input into two parts and processes them separately on different
GPUs. */
// Disable some warnings which are caused with CUDA headers
#if defined(_MSC_VER)
#pragma warning(disable: 4201 4408 4100)
#endif
#include <iostream>
#include "cvconfig.h"
#include "opencv2/core/core.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/gpu/gpu.hpp"
#if !defined(HAVE_CUDA) || !defined(HAVE_TBB)
int main()
{
#if !defined(HAVE_CUDA)
std::cout << "CUDA support is required (CMake key 'WITH_CUDA' must be true).\n";
#endif
#if !defined(HAVE_TBB)
std::cout << "TBB support is required (CMake key 'WITH_TBB' must be true).\n";
#endif
return 0;
}
#else
#include <cuda.h>
#include <cuda_runtime.h>
#include "opencv2/core/internal.hpp" // For TBB wrappers
using namespace std;
using namespace cv;
using namespace cv::gpu;
struct Worker { void operator()(int device_id) const; };
void destroyContexts();
#define safeCall(expr) safeCall_(expr, #expr, __FILE__, __LINE__)
inline void safeCall_(int code, const char* expr, const char* file, int line)
{
if (code != CUDA_SUCCESS)
{
std::cout << "CUDA driver API error: code " << code << ", expr " << expr
<< ", file " << file << ", line " << line << endl;
destroyContexts();
exit(-1);
}
}
// Each GPU is associated with its own context
CUcontext contexts[2];
void inline contextOn(int id)
{
safeCall(cuCtxPushCurrent(contexts[id]));
}
void inline contextOff()
{
CUcontext prev_context;
safeCall(cuCtxPopCurrent(&prev_context));
}
// GPUs data
GpuMat d_left[2];
GpuMat d_right[2];
StereoBM_GPU* bm[2];
GpuMat d_result[2];
// CPU result
Mat result;
void printHelp()
{
std::cout << "Usage: driver_api_stereo_multi_gpu --left <left_image> --right <right_image>\n";
}
int main(int argc, char** argv)
{
if (argc < 5)
{
printHelp();
return -1;
}
int num_devices = getCudaEnabledDeviceCount();
if (num_devices < 2)
{
std::cout << "Two or more GPUs are required\n";
return -1;
}
for (int i = 0; i < num_devices; ++i)
{
cv::gpu::printShortCudaDeviceInfo(i);
DeviceInfo dev_info(i);
if (!dev_info.isCompatible())
{
std::cout << "GPU module isn't built for GPU #" << i << " ("
<< dev_info.name() << ", CC " << dev_info.majorVersion()
<< dev_info.minorVersion() << "\n";
return -1;
}
}
// Load input data
Mat left, right;
for (int i = 1; i < argc; ++i)
{
if (string(argv[i]) == "--left")
{
left = imread(argv[++i], CV_LOAD_IMAGE_GRAYSCALE);
CV_Assert(!left.empty());
}
else if (string(argv[i]) == "--right")
{
right = imread(argv[++i], CV_LOAD_IMAGE_GRAYSCALE);
CV_Assert(!right.empty());
}
else if (string(argv[i]) == "--help")
{
printHelp();
return -1;
}
}
// Init CUDA Driver API
safeCall(cuInit(0));
// Create context for GPU #0
CUdevice device;
safeCall(cuDeviceGet(&device, 0));
safeCall(cuCtxCreate(&contexts[0], 0, device));
contextOff();
// Create context for GPU #1
safeCall(cuDeviceGet(&device, 1));
safeCall(cuCtxCreate(&contexts[1], 0, device));
contextOff();
// Split source images for processing on GPU #0
contextOn(0);
d_left[0].upload(left.rowRange(0, left.rows / 2));
d_right[0].upload(right.rowRange(0, right.rows / 2));
bm[0] = new StereoBM_GPU();
contextOff();
// Split source images for processing on the GPU #1
contextOn(1);
d_left[1].upload(left.rowRange(left.rows / 2, left.rows));
d_right[1].upload(right.rowRange(right.rows / 2, right.rows));
bm[1] = new StereoBM_GPU();
contextOff();
// Execute calculation in two threads using two GPUs
int devices[] = {0, 1};
parallel_do(devices, devices + 2, Worker());
// Release the first GPU resources
contextOn(0);
imshow("GPU #0 result", Mat(d_result[0]));
d_left[0].release();
d_right[0].release();
d_result[0].release();
delete bm[0];
contextOff();
// Release the second GPU resources
contextOn(1);
imshow("GPU #1 result", Mat(d_result[1]));
d_left[1].release();
d_right[1].release();
d_result[1].release();
delete bm[1];
contextOff();
waitKey();
destroyContexts();
return 0;
}
void Worker::operator()(int device_id) const
{
contextOn(device_id);
bm[device_id]->operator()(d_left[device_id], d_right[device_id],
d_result[device_id]);
std::cout << "GPU #" << device_id << " (" << DeviceInfo().name()
<< "): finished\n";
contextOff();
}
void destroyContexts()
{
safeCall(cuCtxDestroy(contexts[0]));
safeCall(cuCtxDestroy(contexts[1]));
}
#endif

270
samples/gpu/highgui_gpu.cpp
View File

@@ -1,135 +1,135 @@
#include <iostream>
#include <string>
#include "opencv2/core/core.hpp"
#include "opencv2/core/gpumat.hpp"
#include "opencv2/core/opengl_interop.hpp"
#include "opencv2/gpu/gpu.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/contrib/contrib.hpp"
using namespace std;
using namespace cv;
using namespace cv::gpu;
struct Timer
{
Timer(const string& msg_)
{
msg = msg_;
tm.reset();
tm.start();
}
~Timer()
{
tm.stop();
cout << msg << " " << tm.getTimeMilli() << " ms\n";
}
string msg;
TickMeter tm;
};
int main(int argc, char* argv[])
{
if (argc < 2)
{
cout << "Usage: " << argv[0] << " image" << endl;
return -1;
}
try
{
bool haveCuda = getCudaEnabledDeviceCount() > 0;
const string openGlMatWnd = "OpenGL Mat";
const string openGlBufferWnd = "OpenGL GlBuffer";
const string openGlTextureWnd = "OpenGL GlTexture";
const string openGlGpuMatWnd = "OpenGL GpuMat";
const string matWnd = "Mat";
namedWindow(openGlMatWnd, WINDOW_OPENGL | WINDOW_AUTOSIZE);
namedWindow(openGlBufferWnd, WINDOW_OPENGL | WINDOW_AUTOSIZE);
namedWindow(openGlTextureWnd, WINDOW_OPENGL | WINDOW_AUTOSIZE);
if (haveCuda)
namedWindow(openGlGpuMatWnd, WINDOW_OPENGL | WINDOW_AUTOSIZE);
namedWindow("Mat", WINDOW_AUTOSIZE);
Mat img = imread(argv[1]);
if (haveCuda)
setGlDevice();
setOpenGlContext(openGlBufferWnd);
GlBuffer buf(img, GlBuffer::TEXTURE_BUFFER);
setOpenGlContext(openGlTextureWnd);
GlTexture tex(img);
GpuMat d_img;
if (haveCuda)
d_img.upload(img);
cout << "=== First call\n\n";
{
Timer t("OpenGL Mat ");
imshow(openGlMatWnd, img);
}
{
Timer t("OpenGL GlBuffer ");
imshow(openGlBufferWnd, buf);
}
{
Timer t("OpenGL GlTexture");
imshow(openGlTextureWnd, tex);
}
if (haveCuda)
{
Timer t("OpenGL GpuMat ");
imshow(openGlGpuMatWnd, d_img);
}
{
Timer t("Mat ");
imshow(matWnd, img);
}
waitKey();
cout << "\n=== Second call\n\n";
{
Timer t("OpenGL Mat ");
imshow(openGlMatWnd, img);
}
{
Timer t("OpenGL GlBuffer ");
imshow(openGlBufferWnd, buf);
}
{
Timer t("OpenGL GlTexture");
imshow(openGlTextureWnd, tex);
}
if (haveCuda)
{
Timer t("OpenGL GpuMat ");
imshow(openGlGpuMatWnd, d_img);
}
{
Timer t("Mat ");
imshow(matWnd, img);
}
cout << "\n";
waitKey();
}
catch(const exception& e)
{
cout << e.what() << endl;
}
return 0;
}
#include <iostream>
#include <string>
#include "opencv2/core/core.hpp"
#include "opencv2/core/gpumat.hpp"
#include "opencv2/core/opengl_interop.hpp"
#include "opencv2/gpu/gpu.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/contrib/contrib.hpp"
using namespace std;
using namespace cv;
using namespace cv::gpu;
struct Timer
{
Timer(const string& msg_)
{
msg = msg_;
tm.reset();
tm.start();
}
~Timer()
{
tm.stop();
cout << msg << " " << tm.getTimeMilli() << " ms\n";
}
string msg;
TickMeter tm;
};
int main(int argc, char* argv[])
{
if (argc < 2)
{
cout << "Usage: " << argv[0] << " image" << endl;
return -1;
}
try
{
bool haveCuda = getCudaEnabledDeviceCount() > 0;
const string openGlMatWnd = "OpenGL Mat";
const string openGlBufferWnd = "OpenGL GlBuffer";
const string openGlTextureWnd = "OpenGL GlTexture";
const string openGlGpuMatWnd = "OpenGL GpuMat";
const string matWnd = "Mat";
namedWindow(openGlMatWnd, WINDOW_OPENGL | WINDOW_AUTOSIZE);
namedWindow(openGlBufferWnd, WINDOW_OPENGL | WINDOW_AUTOSIZE);
namedWindow(openGlTextureWnd, WINDOW_OPENGL | WINDOW_AUTOSIZE);
if (haveCuda)
namedWindow(openGlGpuMatWnd, WINDOW_OPENGL | WINDOW_AUTOSIZE);
namedWindow("Mat", WINDOW_AUTOSIZE);
Mat img = imread(argv[1]);
if (haveCuda)
setGlDevice();
setOpenGlContext(openGlBufferWnd);
GlBuffer buf(img, GlBuffer::TEXTURE_BUFFER);
setOpenGlContext(openGlTextureWnd);
GlTexture tex(img);
GpuMat d_img;
if (haveCuda)
d_img.upload(img);
cout << "=== First call\n\n";
{
Timer t("OpenGL Mat ");
imshow(openGlMatWnd, img);
}
{
Timer t("OpenGL GlBuffer ");
imshow(openGlBufferWnd, buf);
}
{
Timer t("OpenGL GlTexture");
imshow(openGlTextureWnd, tex);
}
if (haveCuda)
{
Timer t("OpenGL GpuMat ");
imshow(openGlGpuMatWnd, d_img);
}
{
Timer t("Mat ");
imshow(matWnd, img);
}
waitKey();
cout << "\n=== Second call\n\n";
{
Timer t("OpenGL Mat ");
imshow(openGlMatWnd, img);
}
{
Timer t("OpenGL GlBuffer ");
imshow(openGlBufferWnd, buf);
}
{
Timer t("OpenGL GlTexture");
imshow(openGlTextureWnd, tex);
}
if (haveCuda)
{
Timer t("OpenGL GpuMat ");
imshow(openGlGpuMatWnd, d_img);
}
{
Timer t("Mat ");
imshow(matWnd, img);
}
cout << "\n";
waitKey();
}
catch(const exception& e)
{
cout << e.what() << endl;
}
return 0;
}

918
samples/gpu/hog.cpp
View File

@@ -1,459 +1,459 @@
#include <iostream>
#include <fstream>
#include <string>
#include <sstream>
#include <iomanip>
#include <stdexcept>
#include "opencv2/gpu/gpu.hpp"
#include "opencv2/highgui/highgui.hpp"
using namespace std;
using namespace cv;
bool help_showed = false;
class Args
{
public:
Args();
static Args read(int argc, char** argv);
string src;
bool src_is_video;
bool src_is_camera;
int camera_id;
bool write_video;
string dst_video;
double dst_video_fps;
bool make_gray;
bool resize_src;
int width, height;
double scale;
int nlevels;
int gr_threshold;
double hit_threshold;
bool hit_threshold_auto;
int win_width;
int win_stride_width, win_stride_height;
bool gamma_corr;
};
class App
{
public:
App(const Args& s);
void run();
void handleKey(char key);
void hogWorkBegin();
void hogWorkEnd();
string hogWorkFps() const;
void workBegin();
void workEnd();
string workFps() const;
string message() const;
private:
App operator=(App&);
Args args;
bool running;
bool use_gpu;
bool make_gray;
double scale;
int gr_threshold;
int nlevels;
double hit_threshold;
bool gamma_corr;
int64 hog_work_begin;
double hog_work_fps;
int64 work_begin;
double work_fps;
};
static void printHelp()
{
cout << "Histogram of Oriented Gradients descriptor and detector sample.\n"
<< "\nUsage: hog_gpu\n"
<< " (<image>|--video <vide>|--camera <camera_id>) # frames source\n"
<< " [--make_gray <true/false>] # convert image to gray one or not\n"
<< " [--resize_src <true/false>] # do resize of the source image or not\n"
<< " [--width <int>] # resized image width\n"
<< " [--height <int>] # resized image height\n"
<< " [--hit_threshold <double>] # classifying plane distance threshold (0.0 usually)\n"
<< " [--scale <double>] # HOG window scale factor\n"
<< " [--nlevels <int>] # max number of HOG window scales\n"
<< " [--win_width <int>] # width of the window (48 or 64)\n"
<< " [--win_stride_width <int>] # distance by OX axis between neighbour wins\n"
<< " [--win_stride_height <int>] # distance by OY axis between neighbour wins\n"
<< " [--gr_threshold <int>] # merging similar rects constant\n"
<< " [--gamma_correct <int>] # do gamma correction or not\n"
<< " [--write_video <bool>] # write video or not\n"
<< " [--dst_video <path>] # output video path\n"
<< " [--dst_video_fps <double>] # output video fps\n";
help_showed = true;
}
int main(int argc, char** argv)
{
try
{
if (argc < 2)
printHelp();
Args args = Args::read(argc, argv);
if (help_showed)
return -1;
App app(args);
app.run();
}
catch (const Exception& e) { return cout << "error: " << e.what() << endl, 1; }
catch (const exception& e) { return cout << "error: " << e.what() << endl, 1; }
catch(...) { return cout << "unknown exception" << endl, 1; }
return 0;
}
Args::Args()
{
src_is_video = false;
src_is_camera = false;
camera_id = 0;
write_video = false;
dst_video_fps = 24.;
make_gray = false;
resize_src = false;
width = 640;
height = 480;
scale = 1.05;
nlevels = 13;
gr_threshold = 8;
hit_threshold = 1.4;
hit_threshold_auto = true;
win_width = 48;
win_stride_width = 8;
win_stride_height = 8;
gamma_corr = true;
}
Args Args::read(int argc, char** argv)
{
Args args;
for (int i = 1; i < argc; i++)
{
if (string(argv[i]) == "--make_gray") args.make_gray = (string(argv[++i]) == "true");
else if (string(argv[i]) == "--resize_src") args.resize_src = (string(argv[++i]) == "true");
else if (string(argv[i]) == "--width") args.width = atoi(argv[++i]);
else if (string(argv[i]) == "--height") args.height = atoi(argv[++i]);
else if (string(argv[i]) == "--hit_threshold")
{
args.hit_threshold = atof(argv[++i]);
args.hit_threshold_auto = false;
}
else if (string(argv[i]) == "--scale") args.scale = atof(argv[++i]);
else if (string(argv[i]) == "--nlevels") args.nlevels = atoi(argv[++i]);
else if (string(argv[i]) == "--win_width") args.win_width = atoi(argv[++i]);
else if (string(argv[i]) == "--win_stride_width") args.win_stride_width = atoi(argv[++i]);
else if (string(argv[i]) == "--win_stride_height") args.win_stride_height = atoi(argv[++i]);
else if (string(argv[i]) == "--gr_threshold") args.gr_threshold = atoi(argv[++i]);
else if (string(argv[i]) == "--gamma_correct") args.gamma_corr = (string(argv[++i]) == "true");
else if (string(argv[i]) == "--write_video") args.write_video = (string(argv[++i]) == "true");
else if (string(argv[i]) == "--dst_video") args.dst_video = argv[++i];
else if (string(argv[i]) == "--dst_video_fps") args.dst_video_fps = atof(argv[++i]);
else if (string(argv[i]) == "--help") printHelp();
else if (string(argv[i]) == "--video") { args.src = argv[++i]; args.src_is_video = true; }
else if (string(argv[i]) == "--camera") { args.camera_id = atoi(argv[++i]); args.src_is_camera = true; }
else if (args.src.empty()) args.src = argv[i];
else throw runtime_error((string("unknown key: ") + argv[i]));
}
return args;
}
App::App(const Args& s)
{
cv::gpu::printShortCudaDeviceInfo(cv::gpu::getDevice());
args = s;
cout << "\nControls:\n"
<< "\tESC - exit\n"
<< "\tm - change mode GPU <-> CPU\n"
<< "\tg - convert image to gray or not\n"
<< "\t1/q - increase/decrease HOG scale\n"
<< "\t2/w - increase/decrease levels count\n"
<< "\t3/e - increase/decrease HOG group threshold\n"
<< "\t4/r - increase/decrease hit threshold\n"
<< endl;
use_gpu = true;
make_gray = args.make_gray;
scale = args.scale;
gr_threshold = args.gr_threshold;
nlevels = args.nlevels;
if (args.hit_threshold_auto)
args.hit_threshold = args.win_width == 48 ? 1.4 : 0.;
hit_threshold = args.hit_threshold;
gamma_corr = args.gamma_corr;
if (args.win_width != 64 && args.win_width != 48)
args.win_width = 64;
cout << "Scale: " << scale << endl;
if (args.resize_src)
cout << "Resized source: (" << args.width << ", " << args.height << ")\n";
cout << "Group threshold: " << gr_threshold << endl;
cout << "Levels number: " << nlevels << endl;
cout << "Win width: " << args.win_width << endl;
cout << "Win stride: (" << args.win_stride_width << ", " << args.win_stride_height << ")\n";
cout << "Hit threshold: " << hit_threshold << endl;
cout << "Gamma correction: " << gamma_corr << endl;
cout << endl;
}
void App::run()
{
running = true;
cv::VideoWriter video_writer;
Size win_size(args.win_width, args.win_width * 2); //(64, 128) or (48, 96)
Size win_stride(args.win_stride_width, args.win_stride_height);
// Create HOG descriptors and detectors here
vector<float> detector;
if (win_size == Size(64, 128))
detector = cv::gpu::HOGDescriptor::getPeopleDetector64x128();
else
detector = cv::gpu::HOGDescriptor::getPeopleDetector48x96();
cv::gpu::HOGDescriptor gpu_hog(win_size, Size(16, 16), Size(8, 8), Size(8, 8), 9,
cv::gpu::HOGDescriptor::DEFAULT_WIN_SIGMA, 0.2, gamma_corr,
cv::gpu::HOGDescriptor::DEFAULT_NLEVELS);
cv::HOGDescriptor cpu_hog(win_size, Size(16, 16), Size(8, 8), Size(8, 8), 9, 1, -1,
HOGDescriptor::L2Hys, 0.2, gamma_corr, cv::HOGDescriptor::DEFAULT_NLEVELS);
gpu_hog.setSVMDetector(detector);
cpu_hog.setSVMDetector(detector);
while (running)
{
VideoCapture vc;
Mat frame;
if (args.src_is_video)
{
vc.open(args.src.c_str());
if (!vc.isOpened())
throw runtime_error(string("can't open video file: " + args.src));
vc >> frame;
}
else if (args.src_is_camera)
{
vc.open(args.camera_id);
if (!vc.isOpened())
{
stringstream msg;
msg << "can't open camera: " << args.camera_id;
throw runtime_error(msg.str());
}
vc >> frame;
}
else
{
frame = imread(args.src);
if (frame.empty())
throw runtime_error(string("can't open image file: " + args.src));
}
Mat img_aux, img, img_to_show;
gpu::GpuMat gpu_img;
// Iterate over all frames
while (running && !frame.empty())
{
workBegin();
// Change format of the image
if (make_gray) cvtColor(frame, img_aux, CV_BGR2GRAY);
else if (use_gpu) cvtColor(frame, img_aux, CV_BGR2BGRA);
else frame.copyTo(img_aux);
// Resize image
if (args.resize_src) resize(img_aux, img, Size(args.width, args.height));
else img = img_aux;
img_to_show = img;
gpu_hog.nlevels = nlevels;
cpu_hog.nlevels = nlevels;
vector<Rect> found;
// Perform HOG classification
hogWorkBegin();
if (use_gpu)
{
gpu_img.upload(img);
gpu_hog.detectMultiScale(gpu_img, found, hit_threshold, win_stride,
Size(0, 0), scale, gr_threshold);
}
else cpu_hog.detectMultiScale(img, found, hit_threshold, win_stride,
Size(0, 0), scale, gr_threshold);
hogWorkEnd();
// Draw positive classified windows
for (size_t i = 0; i < found.size(); i++)
{
Rect r = found[i];
rectangle(img_to_show, r.tl(), r.br(), CV_RGB(0, 255, 0), 3);
}
if (use_gpu)
putText(img_to_show, "Mode: GPU", Point(5, 25), FONT_HERSHEY_SIMPLEX, 1., Scalar(255, 100, 0), 2);
else
putText(img_to_show, "Mode: CPU", Point(5, 25), FONT_HERSHEY_SIMPLEX, 1., Scalar(255, 100, 0), 2);
putText(img_to_show, "FPS (HOG only): " + hogWorkFps(), Point(5, 65), FONT_HERSHEY_SIMPLEX, 1., Scalar(255, 100, 0), 2);
putText(img_to_show, "FPS (total): " + workFps(), Point(5, 105), FONT_HERSHEY_SIMPLEX, 1., Scalar(255, 100, 0), 2);
imshow("opencv_gpu_hog", img_to_show);
if (args.src_is_video || args.src_is_camera) vc >> frame;
workEnd();
if (args.write_video)
{
if (!video_writer.isOpened())
{
video_writer.open(args.dst_video, CV_FOURCC('x','v','i','d'), args.dst_video_fps,
img_to_show.size(), true);
if (!video_writer.isOpened())
throw std::runtime_error("can't create video writer");
}
if (make_gray) cvtColor(img_to_show, img, CV_GRAY2BGR);
else cvtColor(img_to_show, img, CV_BGRA2BGR);
video_writer << img;
}
handleKey((char)waitKey(3));
}
}
}
void App::handleKey(char key)
{
switch (key)
{
case 27:
running = false;
break;
case 'm':
case 'M':
use_gpu = !use_gpu;
cout << "Switched to " << (use_gpu ? "CUDA" : "CPU") << " mode\n";
break;
case 'g':
case 'G':
make_gray = !make_gray;
cout << "Convert image to gray: " << (make_gray ? "YES" : "NO") << endl;
break;
case '1':
scale *= 1.05;
cout << "Scale: " << scale << endl;
break;
case 'q':
case 'Q':
scale /= 1.05;
cout << "Scale: " << scale << endl;
break;
case '2':
nlevels++;
cout << "Levels number: " << nlevels << endl;
break;
case 'w':
case 'W':
nlevels = max(nlevels - 1, 1);
cout << "Levels number: " << nlevels << endl;
break;
case '3':
gr_threshold++;
cout << "Group threshold: " << gr_threshold << endl;
break;
case 'e':
case 'E':
gr_threshold = max(0, gr_threshold - 1);
cout << "Group threshold: " << gr_threshold << endl;
break;
case '4':
hit_threshold+=0.25;
cout << "Hit threshold: " << hit_threshold << endl;
break;
case 'r':
case 'R':
hit_threshold = max(0.0, hit_threshold - 0.25);
cout << "Hit threshold: " << hit_threshold << endl;
break;
case 'c':
case 'C':
gamma_corr = !gamma_corr;
cout << "Gamma correction: " << gamma_corr << endl;
break;
}
}
inline void App::hogWorkBegin() { hog_work_begin = getTickCount(); }
inline void App::hogWorkEnd()
{
int64 delta = getTickCount() - hog_work_begin;
double freq = getTickFrequency();
hog_work_fps = freq / delta;
}
inline string App::hogWorkFps() const
{
stringstream ss;
ss << hog_work_fps;
return ss.str();
}
inline void App::workBegin() { work_begin = getTickCount(); }
inline void App::workEnd()
{
int64 delta = getTickCount() - work_begin;
double freq = getTickFrequency();
work_fps = freq / delta;
}
inline string App::workFps() const
{
stringstream ss;
ss << work_fps;
return ss.str();
}
#include <iostream>
#include <fstream>
#include <string>
#include <sstream>
#include <iomanip>
#include <stdexcept>
#include "opencv2/gpu/gpu.hpp"
#include "opencv2/highgui/highgui.hpp"
using namespace std;
using namespace cv;
bool help_showed = false;
class Args
{
public:
Args();
static Args read(int argc, char** argv);
string src;
bool src_is_video;
bool src_is_camera;
int camera_id;
bool write_video;
string dst_video;
double dst_video_fps;
bool make_gray;
bool resize_src;
int width, height;
double scale;
int nlevels;
int gr_threshold;
double hit_threshold;
bool hit_threshold_auto;
int win_width;
int win_stride_width, win_stride_height;
bool gamma_corr;
};
class App
{
public:
App(const Args& s);
void run();
void handleKey(char key);
void hogWorkBegin();
void hogWorkEnd();
string hogWorkFps() const;
void workBegin();
void workEnd();
string workFps() const;
string message() const;
private:
App operator=(App&);
Args args;
bool running;
bool use_gpu;
bool make_gray;
double scale;
int gr_threshold;
int nlevels;
double hit_threshold;
bool gamma_corr;
int64 hog_work_begin;
double hog_work_fps;
int64 work_begin;
double work_fps;
};
static void printHelp()
{
cout << "Histogram of Oriented Gradients descriptor and detector sample.\n"
<< "\nUsage: hog_gpu\n"
<< " (<image>|--video <vide>|--camera <camera_id>) # frames source\n"
<< " [--make_gray <true/false>] # convert image to gray one or not\n"
<< " [--resize_src <true/false>] # do resize of the source image or not\n"
<< " [--width <int>] # resized image width\n"
<< " [--height <int>] # resized image height\n"
<< " [--hit_threshold <double>] # classifying plane distance threshold (0.0 usually)\n"
<< " [--scale <double>] # HOG window scale factor\n"
<< " [--nlevels <int>] # max number of HOG window scales\n"
<< " [--win_width <int>] # width of the window (48 or 64)\n"
<< " [--win_stride_width <int>] # distance by OX axis between neighbour wins\n"
<< " [--win_stride_height <int>] # distance by OY axis between neighbour wins\n"
<< " [--gr_threshold <int>] # merging similar rects constant\n"
<< " [--gamma_correct <int>] # do gamma correction or not\n"
<< " [--write_video <bool>] # write video or not\n"
<< " [--dst_video <path>] # output video path\n"
<< " [--dst_video_fps <double>] # output video fps\n";
help_showed = true;
}
int main(int argc, char** argv)
{
try
{
if (argc < 2)
printHelp();
Args args = Args::read(argc, argv);
if (help_showed)
return -1;
App app(args);
app.run();
}
catch (const Exception& e) { return cout << "error: " << e.what() << endl, 1; }
catch (const exception& e) { return cout << "error: " << e.what() << endl, 1; }
catch(...) { return cout << "unknown exception" << endl, 1; }
return 0;
}
Args::Args()
{
src_is_video = false;
src_is_camera = false;
camera_id = 0;
write_video = false;
dst_video_fps = 24.;
make_gray = false;
resize_src = false;
width = 640;
height = 480;
scale = 1.05;
nlevels = 13;
gr_threshold = 8;
hit_threshold = 1.4;
hit_threshold_auto = true;
win_width = 48;
win_stride_width = 8;
win_stride_height = 8;
gamma_corr = true;
}
Args Args::read(int argc, char** argv)
{
Args args;
for (int i = 1; i < argc; i++)
{
if (string(argv[i]) == "--make_gray") args.make_gray = (string(argv[++i]) == "true");
else if (string(argv[i]) == "--resize_src") args.resize_src = (string(argv[++i]) == "true");
else if (string(argv[i]) == "--width") args.width = atoi(argv[++i]);
else if (string(argv[i]) == "--height") args.height = atoi(argv[++i]);
else if (string(argv[i]) == "--hit_threshold")
{
args.hit_threshold = atof(argv[++i]);
args.hit_threshold_auto = false;
}
else if (string(argv[i]) == "--scale") args.scale = atof(argv[++i]);
else if (string(argv[i]) == "--nlevels") args.nlevels = atoi(argv[++i]);
else if (string(argv[i]) == "--win_width") args.win_width = atoi(argv[++i]);
else if (string(argv[i]) == "--win_stride_width") args.win_stride_width = atoi(argv[++i]);
else if (string(argv[i]) == "--win_stride_height") args.win_stride_height = atoi(argv[++i]);
else if (string(argv[i]) == "--gr_threshold") args.gr_threshold = atoi(argv[++i]);
else if (string(argv[i]) == "--gamma_correct") args.gamma_corr = (string(argv[++i]) == "true");
else if (string(argv[i]) == "--write_video") args.write_video = (string(argv[++i]) == "true");
else if (string(argv[i]) == "--dst_video") args.dst_video = argv[++i];
else if (string(argv[i]) == "--dst_video_fps") args.dst_video_fps = atof(argv[++i]);
else if (string(argv[i]) == "--help") printHelp();
else if (string(argv[i]) == "--video") { args.src = argv[++i]; args.src_is_video = true; }
else if (string(argv[i]) == "--camera") { args.camera_id = atoi(argv[++i]); args.src_is_camera = true; }
else if (args.src.empty()) args.src = argv[i];
else throw runtime_error((string("unknown key: ") + argv[i]));
}
return args;
}
App::App(const Args& s)
{
cv::gpu::printShortCudaDeviceInfo(cv::gpu::getDevice());
args = s;
cout << "\nControls:\n"
<< "\tESC - exit\n"
<< "\tm - change mode GPU <-> CPU\n"
<< "\tg - convert image to gray or not\n"
<< "\t1/q - increase/decrease HOG scale\n"
<< "\t2/w - increase/decrease levels count\n"
<< "\t3/e - increase/decrease HOG group threshold\n"
<< "\t4/r - increase/decrease hit threshold\n"
<< endl;
use_gpu = true;
make_gray = args.make_gray;
scale = args.scale;
gr_threshold = args.gr_threshold;
nlevels = args.nlevels;
if (args.hit_threshold_auto)
args.hit_threshold = args.win_width == 48 ? 1.4 : 0.;
hit_threshold = args.hit_threshold;
gamma_corr = args.gamma_corr;
if (args.win_width != 64 && args.win_width != 48)
args.win_width = 64;
cout << "Scale: " << scale << endl;
if (args.resize_src)
cout << "Resized source: (" << args.width << ", " << args.height << ")\n";
cout << "Group threshold: " << gr_threshold << endl;
cout << "Levels number: " << nlevels << endl;
cout << "Win width: " << args.win_width << endl;
cout << "Win stride: (" << args.win_stride_width << ", " << args.win_stride_height << ")\n";
cout << "Hit threshold: " << hit_threshold << endl;
cout << "Gamma correction: " << gamma_corr << endl;
cout << endl;
}
void App::run()
{
running = true;
cv::VideoWriter video_writer;
Size win_size(args.win_width, args.win_width * 2); //(64, 128) or (48, 96)
Size win_stride(args.win_stride_width, args.win_stride_height);
// Create HOG descriptors and detectors here
vector<float> detector;
if (win_size == Size(64, 128))
detector = cv::gpu::HOGDescriptor::getPeopleDetector64x128();
else
detector = cv::gpu::HOGDescriptor::getPeopleDetector48x96();
cv::gpu::HOGDescriptor gpu_hog(win_size, Size(16, 16), Size(8, 8), Size(8, 8), 9,
cv::gpu::HOGDescriptor::DEFAULT_WIN_SIGMA, 0.2, gamma_corr,
cv::gpu::HOGDescriptor::DEFAULT_NLEVELS);
cv::HOGDescriptor cpu_hog(win_size, Size(16, 16), Size(8, 8), Size(8, 8), 9, 1, -1,
HOGDescriptor::L2Hys, 0.2, gamma_corr, cv::HOGDescriptor::DEFAULT_NLEVELS);
gpu_hog.setSVMDetector(detector);
cpu_hog.setSVMDetector(detector);
while (running)
{
VideoCapture vc;
Mat frame;
if (args.src_is_video)
{
vc.open(args.src.c_str());
if (!vc.isOpened())
throw runtime_error(string("can't open video file: " + args.src));
vc >> frame;
}
else if (args.src_is_camera)
{
vc.open(args.camera_id);
if (!vc.isOpened())
{
stringstream msg;
msg << "can't open camera: " << args.camera_id;
throw runtime_error(msg.str());
}
vc >> frame;
}
else
{
frame = imread(args.src);
if (frame.empty())
throw runtime_error(string("can't open image file: " + args.src));
}
Mat img_aux, img, img_to_show;
gpu::GpuMat gpu_img;
// Iterate over all frames
while (running && !frame.empty())
{
workBegin();
// Change format of the image
if (make_gray) cvtColor(frame, img_aux, CV_BGR2GRAY);
else if (use_gpu) cvtColor(frame, img_aux, CV_BGR2BGRA);
else frame.copyTo(img_aux);
// Resize image
if (args.resize_src) resize(img_aux, img, Size(args.width, args.height));
else img = img_aux;
img_to_show = img;
gpu_hog.nlevels = nlevels;
cpu_hog.nlevels = nlevels;
vector<Rect> found;
// Perform HOG classification
hogWorkBegin();
if (use_gpu)
{
gpu_img.upload(img);
gpu_hog.detectMultiScale(gpu_img, found, hit_threshold, win_stride,
Size(0, 0), scale, gr_threshold);
}
else cpu_hog.detectMultiScale(img, found, hit_threshold, win_stride,
Size(0, 0), scale, gr_threshold);
hogWorkEnd();
// Draw positive classified windows
for (size_t i = 0; i < found.size(); i++)
{
Rect r = found[i];
rectangle(img_to_show, r.tl(), r.br(), CV_RGB(0, 255, 0), 3);
}
if (use_gpu)
putText(img_to_show, "Mode: GPU", Point(5, 25), FONT_HERSHEY_SIMPLEX, 1., Scalar(255, 100, 0), 2);
else
putText(img_to_show, "Mode: CPU", Point(5, 25), FONT_HERSHEY_SIMPLEX, 1., Scalar(255, 100, 0), 2);
putText(img_to_show, "FPS (HOG only): " + hogWorkFps(), Point(5, 65), FONT_HERSHEY_SIMPLEX, 1., Scalar(255, 100, 0), 2);
putText(img_to_show, "FPS (total): " + workFps(), Point(5, 105), FONT_HERSHEY_SIMPLEX, 1., Scalar(255, 100, 0), 2);
imshow("opencv_gpu_hog", img_to_show);
if (args.src_is_video || args.src_is_camera) vc >> frame;
workEnd();
if (args.write_video)
{
if (!video_writer.isOpened())
{
video_writer.open(args.dst_video, CV_FOURCC('x','v','i','d'), args.dst_video_fps,
img_to_show.size(), true);
if (!video_writer.isOpened())
throw std::runtime_error("can't create video writer");
}
if (make_gray) cvtColor(img_to_show, img, CV_GRAY2BGR);
else cvtColor(img_to_show, img, CV_BGRA2BGR);
video_writer << img;
}
handleKey((char)waitKey(3));
}
}
}
void App::handleKey(char key)
{
switch (key)
{
case 27:
running = false;
break;
case 'm':
case 'M':
use_gpu = !use_gpu;
cout << "Switched to " << (use_gpu ? "CUDA" : "CPU") << " mode\n";
break;
case 'g':
case 'G':
make_gray = !make_gray;
cout << "Convert image to gray: " << (make_gray ? "YES" : "NO") << endl;
break;
case '1':
scale *= 1.05;
cout << "Scale: " << scale << endl;
break;
case 'q':
case 'Q':
scale /= 1.05;
cout << "Scale: " << scale << endl;
break;
case '2':
nlevels++;
cout << "Levels number: " << nlevels << endl;
break;
case 'w':
case 'W':
nlevels = max(nlevels - 1, 1);
cout << "Levels number: " << nlevels << endl;
break;
case '3':
gr_threshold++;
cout << "Group threshold: " << gr_threshold << endl;
break;
case 'e':
case 'E':
gr_threshold = max(0, gr_threshold - 1);
cout << "Group threshold: " << gr_threshold << endl;
break;
case '4':
hit_threshold+=0.25;
cout << "Hit threshold: " << hit_threshold << endl;
break;
case 'r':
case 'R':
hit_threshold = max(0.0, hit_threshold - 0.25);
cout << "Hit threshold: " << hit_threshold << endl;
break;
case 'c':
case 'C':
gamma_corr = !gamma_corr;
cout << "Gamma correction: " << gamma_corr << endl;
break;
}
}
inline void App::hogWorkBegin() { hog_work_begin = getTickCount(); }
inline void App::hogWorkEnd()
{
int64 delta = getTickCount() - hog_work_begin;
double freq = getTickFrequency();
hog_work_fps = freq / delta;
}
inline string App::hogWorkFps() const
{
stringstream ss;
ss << hog_work_fps;
return ss.str();
}
inline void App::workBegin() { work_begin = getTickCount(); }
inline void App::workEnd()
{
int64 delta = getTickCount() - work_begin;
double freq = getTickFrequency();
work_fps = freq / delta;
}
inline string App::workFps() const
{
stringstream ss;
ss << work_fps;
return ss.str();
}

238
samples/gpu/morfology.cpp
View File

@@ -1,119 +1,119 @@
#include "opencv2/imgproc/imgproc.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/gpu/gpu.hpp"
#include <stdlib.h>
#include <stdio.h>
using namespace cv;
using namespace cv::gpu;
static void help()
{
printf("\nShow off image morphology: erosion, dialation, open and close\n"
"Call:\n morphology2 [image]\n"
"This program also shows use of rect, elipse and cross kernels\n\n");
printf( "Hot keys: \n"
"\tESC - quit the program\n"
"\tr - use rectangle structuring element\n"
"\te - use elliptic structuring element\n"
"\tc - use cross-shaped structuring element\n"
"\tSPACE - loop through all the options\n" );
}
GpuMat src, dst;
int element_shape = MORPH_RECT;
//the address of variable which receives trackbar position update
int max_iters = 10;
int open_close_pos = 0;
int erode_dilate_pos = 0;
// callback function for open/close trackbar
static void OpenClose(int, void*)
{
int n = open_close_pos - max_iters;
int an = n > 0 ? n : -n;
Mat element = getStructuringElement(element_shape, Size(an*2+1, an*2+1), Point(an, an) );
if( n < 0 )
cv::gpu::morphologyEx(src, dst, CV_MOP_OPEN, element);
else
cv::gpu::morphologyEx(src, dst, CV_MOP_CLOSE, element);
imshow("Open/Close",(Mat)dst);
}
// callback function for erode/dilate trackbar
static void ErodeDilate(int, void*)
{
int n = erode_dilate_pos - max_iters;
int an = n > 0 ? n : -n;
Mat element = getStructuringElement(element_shape, Size(an*2+1, an*2+1), Point(an, an) );
if( n < 0 )
cv::gpu::erode(src, dst, element);
else
cv::gpu::dilate(src, dst, element);
imshow("Erode/Dilate",(Mat)dst);
}
int main( int argc, char** argv )
{
char* filename = argc == 2 ? argv[1] : (char*)"baboon.jpg";
if (string(argv[1]) == "--help")
{
help();
return -1;
}
src.upload(imread(filename, 1));
if (src.empty())
{
help();
return -1;
}
cv::gpu::printShortCudaDeviceInfo(cv::gpu::getDevice());
help();
if (src.channels() == 3)
{
// gpu support only 4th channel images
GpuMat src4ch;
cv::gpu::cvtColor(src, src4ch, CV_BGR2BGRA);
src = src4ch;
}
//create windows for output images
namedWindow("Open/Close",1);
namedWindow("Erode/Dilate",1);
open_close_pos = erode_dilate_pos = max_iters;
createTrackbar("iterations", "Open/Close",&open_close_pos,max_iters*2+1,OpenClose);
createTrackbar("iterations", "Erode/Dilate",&erode_dilate_pos,max_iters*2+1,ErodeDilate);
for(;;)
{
int c;
OpenClose(open_close_pos, 0);
ErodeDilate(erode_dilate_pos, 0);
c = cvWaitKey(0);
if( (char)c == 27 )
break;
if( (char)c == 'e' )
element_shape = MORPH_ELLIPSE;
else if( (char)c == 'r' )
element_shape = MORPH_RECT;
else if( (char)c == 'c' )
element_shape = MORPH_CROSS;
else if( (char)c == ' ' )
element_shape = (element_shape + 1) % 3;
}
return 0;
}
#include "opencv2/imgproc/imgproc.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/gpu/gpu.hpp"
#include <stdlib.h>
#include <stdio.h>
using namespace cv;
using namespace cv::gpu;
static void help()
{
printf("\nShow off image morphology: erosion, dialation, open and close\n"
"Call:\n morphology2 [image]\n"
"This program also shows use of rect, elipse and cross kernels\n\n");
printf( "Hot keys: \n"
"\tESC - quit the program\n"
"\tr - use rectangle structuring element\n"
"\te - use elliptic structuring element\n"
"\tc - use cross-shaped structuring element\n"
"\tSPACE - loop through all the options\n" );
}
GpuMat src, dst;
int element_shape = MORPH_RECT;
//the address of variable which receives trackbar position update
int max_iters = 10;
int open_close_pos = 0;
int erode_dilate_pos = 0;
// callback function for open/close trackbar
static void OpenClose(int, void*)
{
int n = open_close_pos - max_iters;
int an = n > 0 ? n : -n;
Mat element = getStructuringElement(element_shape, Size(an*2+1, an*2+1), Point(an, an) );
if( n < 0 )
cv::gpu::morphologyEx(src, dst, CV_MOP_OPEN, element);
else
cv::gpu::morphologyEx(src, dst, CV_MOP_CLOSE, element);
imshow("Open/Close",(Mat)dst);
}
// callback function for erode/dilate trackbar
static void ErodeDilate(int, void*)
{
int n = erode_dilate_pos - max_iters;
int an = n > 0 ? n : -n;
Mat element = getStructuringElement(element_shape, Size(an*2+1, an*2+1), Point(an, an) );
if( n < 0 )
cv::gpu::erode(src, dst, element);
else
cv::gpu::dilate(src, dst, element);
imshow("Erode/Dilate",(Mat)dst);
}
int main( int argc, char** argv )
{
char* filename = argc == 2 ? argv[1] : (char*)"baboon.jpg";
if (string(argv[1]) == "--help")
{
help();
return -1;
}
src.upload(imread(filename, 1));
if (src.empty())
{
help();
return -1;
}
cv::gpu::printShortCudaDeviceInfo(cv::gpu::getDevice());
help();
if (src.channels() == 3)
{
// gpu support only 4th channel images
GpuMat src4ch;
cv::gpu::cvtColor(src, src4ch, CV_BGR2BGRA);
src = src4ch;
}
//create windows for output images
namedWindow("Open/Close",1);
namedWindow("Erode/Dilate",1);
open_close_pos = erode_dilate_pos = max_iters;
createTrackbar("iterations", "Open/Close",&open_close_pos,max_iters*2+1,OpenClose);
createTrackbar("iterations", "Erode/Dilate",&erode_dilate_pos,max_iters*2+1,ErodeDilate);
for(;;)
{
int c;
OpenClose(open_close_pos, 0);
ErodeDilate(erode_dilate_pos, 0);
c = cvWaitKey(0);
if( (char)c == 27 )
break;
if( (char)c == 'e' )
element_shape = MORPH_ELLIPSE;
else if( (char)c == 'r' )
element_shape = MORPH_RECT;
else if( (char)c == 'c' )
element_shape = MORPH_CROSS;
else if( (char)c == ' ' )
element_shape = (element_shape + 1) % 3;
}
return 0;
}

196
samples/gpu/multi.cpp
View File

@@ -1,98 +1,98 @@
/* This sample demonstrates the way you can perform independed tasks
on the different GPUs */
// Disable some warnings which are caused with CUDA headers
#if defined(_MSC_VER)
#pragma warning(disable: 4201 4408 4100)
#endif
#include <iostream>
#include "cvconfig.h"
#include "opencv2/core/core.hpp"
#include "opencv2/gpu/gpu.hpp"
#if !defined(HAVE_CUDA) || !defined(HAVE_TBB)
int main()
{
#if !defined(HAVE_CUDA)
std::cout << "CUDA support is required (CMake key 'WITH_CUDA' must be true).\n";
#endif
#if !defined(HAVE_TBB)
std::cout << "TBB support is required (CMake key 'WITH_TBB' must be true).\n";
#endif
return 0;
}
#else
#include "opencv2/core/internal.hpp" // For TBB wrappers
using namespace std;
using namespace cv;
using namespace cv::gpu;
struct Worker { void operator()(int device_id) const; };
int main()
{
int num_devices = getCudaEnabledDeviceCount();
if (num_devices < 2)
{
std::cout << "Two or more GPUs are required\n";
return -1;
}
for (int i = 0; i < num_devices; ++i)
{
cv::gpu::printShortCudaDeviceInfo(i);
DeviceInfo dev_info(i);
if (!dev_info.isCompatible())
{
std::cout << "GPU module isn't built for GPU #" << i << " ("
<< dev_info.name() << ", CC " << dev_info.majorVersion()
<< dev_info.minorVersion() << "\n";
return -1;
}
}
// Execute calculation in two threads using two GPUs
int devices[] = {0, 1};
parallel_do(devices, devices + 2, Worker());
return 0;
}
void Worker::operator()(int device_id) const
{
setDevice(device_id);
Mat src(1000, 1000, CV_32F);
Mat dst;
RNG rng(0);
rng.fill(src, RNG::UNIFORM, 0, 1);
// CPU works
transpose(src, dst);
// GPU works
GpuMat d_src(src);
GpuMat d_dst;
transpose(d_src, d_dst);
// Check results
bool passed = norm(dst - Mat(d_dst), NORM_INF) < 1e-3;
std::cout << "GPU #" << device_id << " (" << DeviceInfo().name() << "): "
<< (passed ? "passed" : "FAILED") << endl;
// Deallocate data here, otherwise deallocation will be performed
// after context is extracted from the stack
d_src.release();
d_dst.release();
}
#endif
/* This sample demonstrates the way you can perform independed tasks
on the different GPUs */
// Disable some warnings which are caused with CUDA headers
#if defined(_MSC_VER)
#pragma warning(disable: 4201 4408 4100)
#endif
#include <iostream>
#include "cvconfig.h"
#include "opencv2/core/core.hpp"
#include "opencv2/gpu/gpu.hpp"
#if !defined(HAVE_CUDA) || !defined(HAVE_TBB)
int main()
{
#if !defined(HAVE_CUDA)
std::cout << "CUDA support is required (CMake key 'WITH_CUDA' must be true).\n";
#endif
#if !defined(HAVE_TBB)
std::cout << "TBB support is required (CMake key 'WITH_TBB' must be true).\n";
#endif
return 0;
}
#else
#include "opencv2/core/internal.hpp" // For TBB wrappers
using namespace std;
using namespace cv;
using namespace cv::gpu;
struct Worker { void operator()(int device_id) const; };
int main()
{
int num_devices = getCudaEnabledDeviceCount();
if (num_devices < 2)
{
std::cout << "Two or more GPUs are required\n";
return -1;
}
for (int i = 0; i < num_devices; ++i)
{
cv::gpu::printShortCudaDeviceInfo(i);
DeviceInfo dev_info(i);
if (!dev_info.isCompatible())
{
std::cout << "GPU module isn't built for GPU #" << i << " ("
<< dev_info.name() << ", CC " << dev_info.majorVersion()
<< dev_info.minorVersion() << "\n";
return -1;
}
}
// Execute calculation in two threads using two GPUs
int devices[] = {0, 1};
parallel_do(devices, devices + 2, Worker());
return 0;
}
void Worker::operator()(int device_id) const
{
setDevice(device_id);
Mat src(1000, 1000, CV_32F);
Mat dst;
RNG rng(0);
rng.fill(src, RNG::UNIFORM, 0, 1);
// CPU works
transpose(src, dst);
// GPU works
GpuMat d_src(src);
GpuMat d_dst;
transpose(d_src, d_dst);
// Check results
bool passed = norm(dst - Mat(d_dst), NORM_INF) < 1e-3;
std::cout << "GPU #" << device_id << " (" << DeviceInfo().name() << "): "
<< (passed ? "passed" : "FAILED") << endl;
// Deallocate data here, otherwise deallocation will be performed
// after context is extracted from the stack
d_src.release();
d_dst.release();
}
#endif

1304
samples/gpu/opticalflow_nvidia_api.cpp
View File

File diff suppressed because it is too large Load Diff

52
samples/gpu/performance/CMakeLists.txt
View File

@@ -1,26 +1,26 @@
set(the_target "example_gpu_performance")
file(GLOB sources "performance/*.cpp")
file(GLOB headers "performance/*.h")
add_executable(${the_target} ${sources} ${headers})
target_link_libraries(${the_target} ${OPENCV_LINKER_LIBS} ${OPENCV_GPU_SAMPLES_REQUIRED_DEPS})
set_target_properties(${the_target} PROPERTIES
OUTPUT_NAME "performance_gpu"
PROJECT_LABEL "(EXAMPLE_GPU) performance")
if(ENABLE_SOLUTION_FOLDERS)
set_target_properties(${the_target} PROPERTIES FOLDER "samples//gpu")
endif()
if(WIN32)
install(TARGETS ${the_target} RUNTIME DESTINATION "samples/gpu" COMPONENT main)
endif()
if(INSTALL_C_EXAMPLES AND NOT WIN32)
file(GLOB GPU_FILES performance/*.cpp performance/*.h)
install(FILES ${GPU_FILES}
DESTINATION share/OpenCV/samples/gpu/performance
PERMISSIONS OWNER_READ GROUP_READ WORLD_READ)
endif()
set(the_target "example_gpu_performance")
file(GLOB sources "performance/*.cpp")
file(GLOB headers "performance/*.h")
add_executable(${the_target} ${sources} ${headers})
target_link_libraries(${the_target} ${OPENCV_LINKER_LIBS} ${OPENCV_GPU_SAMPLES_REQUIRED_DEPS})
set_target_properties(${the_target} PROPERTIES
OUTPUT_NAME "performance_gpu"
PROJECT_LABEL "(EXAMPLE_GPU) performance")
if(ENABLE_SOLUTION_FOLDERS)
set_target_properties(${the_target} PROPERTIES FOLDER "samples//gpu")
endif()
if(WIN32)
install(TARGETS ${the_target} RUNTIME DESTINATION "samples/gpu" COMPONENT main)
endif()
if(INSTALL_C_EXAMPLES AND NOT WIN32)
file(GLOB GPU_FILES performance/*.cpp performance/*.h)
install(FILES ${GPU_FILES}
DESTINATION share/OpenCV/samples/gpu/performance
PERMISSIONS OWNER_READ GROUP_READ WORLD_READ)
endif()

452
samples/gpu/performance/performance.cpp
View File

@@ -1,226 +1,226 @@
#include <iomanip>
#include <stdexcept>
#include <string>
#include "performance.h"
using namespace std;
using namespace cv;
using namespace cv::gpu;
void TestSystem::run()
{
if (is_list_mode_)
{
for (vector<Runnable*>::iterator it = tests_.begin(); it != tests_.end(); ++it)
cout << (*it)->name() << endl;
return;
}
// Run test initializers
for (vector<Runnable*>::iterator it = inits_.begin(); it != inits_.end(); ++it)
{
if ((*it)->name().find(test_filter_, 0) != string::npos)
(*it)->run();
}
printHeading();
// Run tests
for (vector<Runnable*>::iterator it = tests_.begin(); it != tests_.end(); ++it)
{
try
{
if ((*it)->name().find(test_filter_, 0) != string::npos)
{
cout << endl << (*it)->name() << ":\n";
(*it)->run();
finishCurrentSubtest();
}
}
catch (const Exception&)
{
// Message is printed via callback
resetCurrentSubtest();
}
catch (const runtime_error& e)
{
printError(e.what());
resetCurrentSubtest();
}
}
printSummary();
}
void TestSystem::finishCurrentSubtest()
{
if (cur_subtest_is_empty_)
// There is no need to print subtest statistics
return;
double cpu_time = cpu_elapsed_ / getTickFrequency() * 1000.0;
double gpu_time = gpu_elapsed_ / getTickFrequency() * 1000.0;
double speedup = static_cast<double>(cpu_elapsed_) / std::max(1.0, gpu_elapsed_);
speedup_total_ += speedup;
printMetrics(cpu_time, gpu_time, speedup);
num_subtests_called_++;
resetCurrentSubtest();
}
double TestSystem::meanTime(const vector<int64> &samples)
{
double sum = accumulate(samples.begin(), samples.end(), 0.);
if (samples.size() > 1)
return (sum - samples[0]) / (samples.size() - 1);
return sum;
}
void TestSystem::printHeading()
{
cout << endl;
cout << setiosflags(ios_base::left);
cout << TAB << setw(10) << "CPU, ms" << setw(10) << "GPU, ms"
<< setw(14) << "SPEEDUP"
<< "DESCRIPTION\n";
cout << resetiosflags(ios_base::left);
}
void TestSystem::printSummary()
{
cout << setiosflags(ios_base::fixed);
cout << "\naverage GPU speedup: x"
<< setprecision(3) << speedup_total_ / std::max(1, num_subtests_called_)
<< endl;
cout << resetiosflags(ios_base::fixed);
}
void TestSystem::printMetrics(double cpu_time, double gpu_time, double speedup)
{
cout << TAB << setiosflags(ios_base::left);
stringstream stream;
stream << cpu_time;
cout << setw(10) << stream.str();
stream.str("");
stream << gpu_time;
cout << setw(10) << stream.str();
stream.str("");
stream << "x" << setprecision(3) << speedup;
cout << setw(14) << stream.str();
cout << cur_subtest_description_.str();
cout << resetiosflags(ios_base::left) << endl;
}
void TestSystem::printError(const std::string& msg)
{
cout << TAB << "[error: " << msg << "] " << cur_subtest_description_.str() << endl;
}
void gen(Mat& mat, int rows, int cols, int type, Scalar low, Scalar high)
{
mat.create(rows, cols, type);
RNG rng(0);
rng.fill(mat, RNG::UNIFORM, low, high);
}
string abspath(const string& relpath)
{
return TestSystem::instance().workingDir() + relpath;
}
static int CV_CDECL cvErrorCallback(int /*status*/, const char* /*func_name*/,
const char* err_msg, const char* /*file_name*/,
int /*line*/, void* /*userdata*/)
{
TestSystem::instance().printError(err_msg);
return 0;
}
int main(int argc, const char* argv[])
{
int num_devices = getCudaEnabledDeviceCount();
if (num_devices == 0)
{
cerr << "No GPU found or the library was compiled without GPU support";
return -1;
}
redirectError(cvErrorCallback);
const char* keys =
"{ h help | | print help message }"
"{ f filter | | filter for test }"
"{ w workdir | | set working directory }"
"{ l list | | show all tests }"
"{ d device | 0 | device id }"
"{ i iters | 10 | iteration count }";
CommandLineParser cmd(argc, argv, keys);
if (cmd.has("help") || !cmd.check())
{
cmd.printMessage();
cmd.printErrors();
return 0;
}
int device = cmd.get<int>("device");
if (device < 0 || device >= num_devices)
{
cerr << "Invalid device ID" << endl;
return -1;
}
DeviceInfo dev_info(device);
if (!dev_info.isCompatible())
{
cerr << "GPU module isn't built for GPU #" << device << " " << dev_info.name() << ", CC " << dev_info.majorVersion() << '.' << dev_info.minorVersion() << endl;
return -1;
}
setDevice(device);
printShortCudaDeviceInfo(device);
string filter = cmd.get<string>("filter");
string workdir = cmd.get<string>("workdir");
bool list = cmd.has("list");
int iters = cmd.get<int>("iters");
if (!filter.empty())
TestSystem::instance().setTestFilter(filter);
if (!workdir.empty())
{
if (workdir[workdir.size() - 1] != '/' && workdir[workdir.size() - 1] != '\\')
workdir += '/';
TestSystem::instance().setWorkingDir(workdir);
}
if (list)
TestSystem::instance().setListMode(true);
TestSystem::instance().setNumIters(iters);
cout << "\nNote: the timings for GPU don't include data transfer" << endl;
TestSystem::instance().run();
return 0;
}
#include <iomanip>
#include <stdexcept>
#include <string>
#include "performance.h"
using namespace std;
using namespace cv;
using namespace cv::gpu;
void TestSystem::run()
{
if (is_list_mode_)
{
for (vector<Runnable*>::iterator it = tests_.begin(); it != tests_.end(); ++it)
cout << (*it)->name() << endl;
return;
}
// Run test initializers
for (vector<Runnable*>::iterator it = inits_.begin(); it != inits_.end(); ++it)
{
if ((*it)->name().find(test_filter_, 0) != string::npos)
(*it)->run();
}
printHeading();
// Run tests
for (vector<Runnable*>::iterator it = tests_.begin(); it != tests_.end(); ++it)
{
try
{
if ((*it)->name().find(test_filter_, 0) != string::npos)
{
cout << endl << (*it)->name() << ":\n";
(*it)->run();
finishCurrentSubtest();
}
}
catch (const Exception&)
{
// Message is printed via callback
resetCurrentSubtest();
}
catch (const runtime_error& e)
{
printError(e.what());
resetCurrentSubtest();
}
}
printSummary();
}
void TestSystem::finishCurrentSubtest()
{
if (cur_subtest_is_empty_)
// There is no need to print subtest statistics
return;
double cpu_time = cpu_elapsed_ / getTickFrequency() * 1000.0;
double gpu_time = gpu_elapsed_ / getTickFrequency() * 1000.0;
double speedup = static_cast<double>(cpu_elapsed_) / std::max(1.0, gpu_elapsed_);
speedup_total_ += speedup;
printMetrics(cpu_time, gpu_time, speedup);
num_subtests_called_++;
resetCurrentSubtest();
}
double TestSystem::meanTime(const vector<int64> &samples)
{
double sum = accumulate(samples.begin(), samples.end(), 0.);
if (samples.size() > 1)
return (sum - samples[0]) / (samples.size() - 1);
return sum;
}
void TestSystem::printHeading()
{
cout << endl;
cout << setiosflags(ios_base::left);
cout << TAB << setw(10) << "CPU, ms" << setw(10) << "GPU, ms"
<< setw(14) << "SPEEDUP"
<< "DESCRIPTION\n";
cout << resetiosflags(ios_base::left);
}
void TestSystem::printSummary()
{
cout << setiosflags(ios_base::fixed);
cout << "\naverage GPU speedup: x"
<< setprecision(3) << speedup_total_ / std::max(1, num_subtests_called_)
<< endl;
cout << resetiosflags(ios_base::fixed);
}
void TestSystem::printMetrics(double cpu_time, double gpu_time, double speedup)
{
cout << TAB << setiosflags(ios_base::left);
stringstream stream;
stream << cpu_time;
cout << setw(10) << stream.str();
stream.str("");
stream << gpu_time;
cout << setw(10) << stream.str();
stream.str("");
stream << "x" << setprecision(3) << speedup;
cout << setw(14) << stream.str();
cout << cur_subtest_description_.str();
cout << resetiosflags(ios_base::left) << endl;
}
void TestSystem::printError(const std::string& msg)
{
cout << TAB << "[error: " << msg << "] " << cur_subtest_description_.str() << endl;
}
void gen(Mat& mat, int rows, int cols, int type, Scalar low, Scalar high)
{
mat.create(rows, cols, type);
RNG rng(0);
rng.fill(mat, RNG::UNIFORM, low, high);
}
string abspath(const string& relpath)
{
return TestSystem::instance().workingDir() + relpath;
}
static int CV_CDECL cvErrorCallback(int /*status*/, const char* /*func_name*/,
const char* err_msg, const char* /*file_name*/,
int /*line*/, void* /*userdata*/)
{
TestSystem::instance().printError(err_msg);
return 0;
}
int main(int argc, const char* argv[])
{
int num_devices = getCudaEnabledDeviceCount();
if (num_devices == 0)
{
cerr << "No GPU found or the library was compiled without GPU support";
return -1;
}
redirectError(cvErrorCallback);
const char* keys =
"{ h help | | print help message }"
"{ f filter | | filter for test }"
"{ w workdir | | set working directory }"
"{ l list | | show all tests }"
"{ d device | 0 | device id }"
"{ i iters | 10 | iteration count }";
CommandLineParser cmd(argc, argv, keys);
if (cmd.has("help") || !cmd.check())
{
cmd.printMessage();
cmd.printErrors();
return 0;
}
int device = cmd.get<int>("device");
if (device < 0 || device >= num_devices)
{
cerr << "Invalid device ID" << endl;
return -1;
}
DeviceInfo dev_info(device);
if (!dev_info.isCompatible())
{
cerr << "GPU module isn't built for GPU #" << device << " " << dev_info.name() << ", CC " << dev_info.majorVersion() << '.' << dev_info.minorVersion() << endl;
return -1;
}
setDevice(device);
printShortCudaDeviceInfo(device);
string filter = cmd.get<string>("filter");
string workdir = cmd.get<string>("workdir");
bool list = cmd.has("list");
int iters = cmd.get<int>("iters");
if (!filter.empty())
TestSystem::instance().setTestFilter(filter);
if (!workdir.empty())
{
if (workdir[workdir.size() - 1] != '/' && workdir[workdir.size() - 1] != '\\')
workdir += '/';
TestSystem::instance().setWorkingDir(workdir);
}
if (list)
TestSystem::instance().setListMode(true);
TestSystem::instance().setNumIters(iters);
cout << "\nNote: the timings for GPU don't include data transfer" << endl;
TestSystem::instance().run();
return 0;
}

378
samples/gpu/performance/performance.h
View File

@@ -1,189 +1,189 @@
#ifndef OPENCV_GPU_SAMPLE_PERFORMANCE_H_
#define OPENCV_GPU_SAMPLE_PERFORMANCE_H_
#include <iostream>
#include <cstdio>
#include <vector>
#include <numeric>
#include <string>
#include "opencv2/core/core.hpp"
#include "opencv2/gpu/gpu.hpp"
#define TAB " "
class Runnable
{
public:
explicit Runnable(const std::string& nameStr): name_(nameStr) {}
virtual ~Runnable() {}
const std::string& name() const { return name_; }
virtual void run() = 0;
private:
std::string name_;
};
class TestSystem
{
public:
static TestSystem& instance()
{
static TestSystem me;
return me;
}
void setWorkingDir(const std::string& val) { working_dir_ = val; }
const std::string& workingDir() const { return working_dir_; }
void setTestFilter(const std::string& val) { test_filter_ = val; }
const std::string& testFilter() const { return test_filter_; }
void setNumIters(int num_iters) { num_iters_ = num_iters; }
void addInit(Runnable* init) { inits_.push_back(init); }
void addTest(Runnable* test) { tests_.push_back(test); }
void run();
// It's public because OpenCV callback uses it
void printError(const std::string& msg);
std::stringstream& startNewSubtest()
{
finishCurrentSubtest();
return cur_subtest_description_;
}
bool stop() const { return cur_iter_idx_ >= num_iters_; }
void cpuOn() { cpu_started_ = cv::getTickCount(); }
void cpuOff()
{
int64 delta = cv::getTickCount() - cpu_started_;
cpu_times_.push_back(delta);
++cur_iter_idx_;
}
void cpuComplete()
{
cpu_elapsed_ += meanTime(cpu_times_);
cur_subtest_is_empty_ = false;
cur_iter_idx_ = 0;
}
void gpuOn() { gpu_started_ = cv::getTickCount(); }
void gpuOff()
{
int64 delta = cv::getTickCount() - gpu_started_;
gpu_times_.push_back(delta);
++cur_iter_idx_;
}
void gpuComplete()
{
gpu_elapsed_ += meanTime(gpu_times_);
cur_subtest_is_empty_ = false;
cur_iter_idx_ = 0;
}
bool isListMode() const { return is_list_mode_; }
void setListMode(bool value) { is_list_mode_ = value; }
private:
TestSystem():
cur_subtest_is_empty_(true), cpu_elapsed_(0),
gpu_elapsed_(0), speedup_total_(0.0),
num_subtests_called_(0), is_list_mode_(false),
num_iters_(10), cur_iter_idx_(0)
{
cpu_times_.reserve(num_iters_);
gpu_times_.reserve(num_iters_);
}
void finishCurrentSubtest();
void resetCurrentSubtest()
{
cpu_elapsed_ = 0;
gpu_elapsed_ = 0;
cur_subtest_description_.str("");
cur_subtest_is_empty_ = true;
cur_iter_idx_ = 0;
cpu_times_.clear();
gpu_times_.clear();
}
double meanTime(const std::vector<int64> &samples);
void printHeading();
void printSummary();
void printMetrics(double cpu_time, double gpu_time, double speedup);
std::string working_dir_;
std::string test_filter_;
std::vector<Runnable*> inits_;
std::vector<Runnable*> tests_;
std::stringstream cur_subtest_description_;
bool cur_subtest_is_empty_;
int64 cpu_started_;
int64 gpu_started_;
double cpu_elapsed_;
double gpu_elapsed_;
double speedup_total_;
int num_subtests_called_;
bool is_list_mode_;
int num_iters_;
int cur_iter_idx_;
std::vector<int64> cpu_times_;
std::vector<int64> gpu_times_;
};
#define GLOBAL_INIT(name) \
struct name##_init: Runnable { \
name##_init(): Runnable(#name) { \
TestSystem::instance().addInit(this); \
} \
void run(); \
} name##_init_instance; \
void name##_init::run()
#define TEST(name) \
struct name##_test: Runnable { \
name##_test(): Runnable(#name) { \
TestSystem::instance().addTest(this); \
} \
void run(); \
} name##_test_instance; \
void name##_test::run()
#define SUBTEST TestSystem::instance().startNewSubtest()
#define CPU_ON \
while (!TestSystem::instance().stop()) { \
TestSystem::instance().cpuOn()
#define CPU_OFF \
TestSystem::instance().cpuOff(); \
} TestSystem::instance().cpuComplete()
#define GPU_ON \
while (!TestSystem::instance().stop()) { \
TestSystem::instance().gpuOn()
#define GPU_OFF \
TestSystem::instance().gpuOff(); \
} TestSystem::instance().gpuComplete()
// Generates a matrix
void gen(cv::Mat& mat, int rows, int cols, int type, cv::Scalar low,
cv::Scalar high);
// Returns abs path taking into account test system working dir
std::string abspath(const std::string& relpath);
#endif // OPENCV_GPU_SAMPLE_PERFORMANCE_H_
#ifndef OPENCV_GPU_SAMPLE_PERFORMANCE_H_
#define OPENCV_GPU_SAMPLE_PERFORMANCE_H_
#include <iostream>
#include <cstdio>
#include <vector>
#include <numeric>
#include <string>
#include "opencv2/core/core.hpp"
#include "opencv2/gpu/gpu.hpp"
#define TAB " "
class Runnable
{
public:
explicit Runnable(const std::string& nameStr): name_(nameStr) {}
virtual ~Runnable() {}
const std::string& name() const { return name_; }
virtual void run() = 0;
private:
std::string name_;
};
class TestSystem
{
public:
static TestSystem& instance()
{
static TestSystem me;
return me;
}
void setWorkingDir(const std::string& val) { working_dir_ = val; }
const std::string& workingDir() const { return working_dir_; }
void setTestFilter(const std::string& val) { test_filter_ = val; }
const std::string& testFilter() const { return test_filter_; }
void setNumIters(int num_iters) { num_iters_ = num_iters; }
void addInit(Runnable* init) { inits_.push_back(init); }
void addTest(Runnable* test) { tests_.push_back(test); }
void run();
// It's public because OpenCV callback uses it
void printError(const std::string& msg);
std::stringstream& startNewSubtest()
{
finishCurrentSubtest();
return cur_subtest_description_;
}
bool stop() const { return cur_iter_idx_ >= num_iters_; }
void cpuOn() { cpu_started_ = cv::getTickCount(); }
void cpuOff()
{
int64 delta = cv::getTickCount() - cpu_started_;
cpu_times_.push_back(delta);
++cur_iter_idx_;
}
void cpuComplete()
{
cpu_elapsed_ += meanTime(cpu_times_);
cur_subtest_is_empty_ = false;
cur_iter_idx_ = 0;
}
void gpuOn() { gpu_started_ = cv::getTickCount(); }
void gpuOff()
{
int64 delta = cv::getTickCount() - gpu_started_;
gpu_times_.push_back(delta);
++cur_iter_idx_;
}
void gpuComplete()
{
gpu_elapsed_ += meanTime(gpu_times_);
cur_subtest_is_empty_ = false;
cur_iter_idx_ = 0;
}
bool isListMode() const { return is_list_mode_; }
void setListMode(bool value) { is_list_mode_ = value; }
private:
TestSystem():
cur_subtest_is_empty_(true), cpu_elapsed_(0),
gpu_elapsed_(0), speedup_total_(0.0),
num_subtests_called_(0), is_list_mode_(false),
num_iters_(10), cur_iter_idx_(0)
{
cpu_times_.reserve(num_iters_);
gpu_times_.reserve(num_iters_);
}
void finishCurrentSubtest();
void resetCurrentSubtest()
{
cpu_elapsed_ = 0;
gpu_elapsed_ = 0;
cur_subtest_description_.str("");
cur_subtest_is_empty_ = true;
cur_iter_idx_ = 0;
cpu_times_.clear();
gpu_times_.clear();
}
double meanTime(const std::vector<int64> &samples);
void printHeading();
void printSummary();
void printMetrics(double cpu_time, double gpu_time, double speedup);
std::string working_dir_;
std::string test_filter_;
std::vector<Runnable*> inits_;
std::vector<Runnable*> tests_;
std::stringstream cur_subtest_description_;
bool cur_subtest_is_empty_;
int64 cpu_started_;
int64 gpu_started_;
double cpu_elapsed_;
double gpu_elapsed_;
double speedup_total_;
int num_subtests_called_;
bool is_list_mode_;
int num_iters_;
int cur_iter_idx_;
std::vector<int64> cpu_times_;
std::vector<int64> gpu_times_;
};
#define GLOBAL_INIT(name) \
struct name##_init: Runnable { \
name##_init(): Runnable(#name) { \
TestSystem::instance().addInit(this); \
} \
void run(); \
} name##_init_instance; \
void name##_init::run()
#define TEST(name) \
struct name##_test: Runnable { \
name##_test(): Runnable(#name) { \
TestSystem::instance().addTest(this); \
} \
void run(); \
} name##_test_instance; \
void name##_test::run()
#define SUBTEST TestSystem::instance().startNewSubtest()
#define CPU_ON \
while (!TestSystem::instance().stop()) { \
TestSystem::instance().cpuOn()
#define CPU_OFF \
TestSystem::instance().cpuOff(); \
} TestSystem::instance().cpuComplete()
#define GPU_ON \
while (!TestSystem::instance().stop()) { \
TestSystem::instance().gpuOn()
#define GPU_OFF \
TestSystem::instance().gpuOff(); \
} TestSystem::instance().gpuComplete()
// Generates a matrix
void gen(cv::Mat& mat, int rows, int cols, int type, cv::Scalar low,
cv::Scalar high);
// Returns abs path taking into account test system working dir
std::string abspath(const std::string& relpath);
#endif // OPENCV_GPU_SAMPLE_PERFORMANCE_H_

2830
samples/gpu/performance/tests.cpp
View File

File diff suppressed because it is too large Load Diff

582
samples/gpu/pyrlk_optical_flow.cpp
View File

@@ -1,291 +1,291 @@
#include <iostream>
#include <vector>
#include "cvconfig.h"
#include "opencv2/core/core.hpp"
#include "opencv2/core/opengl_interop.hpp"
#include "opencv2/imgproc/imgproc.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/video/video.hpp"
#include "opencv2/gpu/gpu.hpp"
using namespace std;
using namespace cv;
using namespace cv::gpu;
static void download(const GpuMat& d_mat, vector<Point2f>& vec)
{
vec.resize(d_mat.cols);
Mat mat(1, d_mat.cols, CV_32FC2, (void*)&vec[0]);
d_mat.download(mat);
}
static void download(const GpuMat& d_mat, vector<uchar>& vec)
{
vec.resize(d_mat.cols);
Mat mat(1, d_mat.cols, CV_8UC1, (void*)&vec[0]);
d_mat.download(mat);
}
static void drawArrows(Mat& frame, const vector<Point2f>& prevPts, const vector<Point2f>& nextPts, const vector<uchar>& status, Scalar line_color = Scalar(0, 0, 255))
{
for (size_t i = 0; i < prevPts.size(); ++i)
{
if (status[i])
{
int line_thickness = 1;
Point p = prevPts[i];
Point q = nextPts[i];
double angle = atan2((double) p.y - q.y, (double) p.x - q.x);
double hypotenuse = sqrt( (double)(p.y - q.y)*(p.y - q.y) + (double)(p.x - q.x)*(p.x - q.x) );
if (hypotenuse < 1.0)
continue;
// Here we lengthen the arrow by a factor of three.
q.x = (int) (p.x - 3 * hypotenuse * cos(angle));
q.y = (int) (p.y - 3 * hypotenuse * sin(angle));
// Now we draw the main line of the arrow.
line(frame, p, q, line_color, line_thickness);
// Now draw the tips of the arrow. I do some scaling so that the
// tips look proportional to the main line of the arrow.
p.x = (int) (q.x + 9 * cos(angle + CV_PI / 4));
p.y = (int) (q.y + 9 * sin(angle + CV_PI / 4));
line(frame, p, q, line_color, line_thickness);
p.x = (int) (q.x + 9 * cos(angle - CV_PI / 4));
p.y = (int) (q.y + 9 * sin(angle - CV_PI / 4));
line(frame, p, q, line_color, line_thickness);
}
}
}
#ifdef HAVE_OPENGL
struct DrawData
{
GlTexture tex;
GlArrays arr;
};
static void drawCallback(void* userdata)
{
DrawData* data = static_cast<DrawData*>(userdata);
if (data->tex.empty() || data->arr.empty())
return;
static GlCamera camera;
static bool init_camera = true;
if (init_camera)
{
camera.setOrthoProjection(0.0, 1.0, 1.0, 0.0, 0.0, 1.0);
camera.lookAt(Point3d(0.0, 0.0, 1.0), Point3d(0.0, 0.0, 0.0), Point3d(0.0, 1.0, 0.0));
init_camera = false;
}
camera.setupProjectionMatrix();
camera.setupModelViewMatrix();
render(data->tex);
render(data->arr, RenderMode::TRIANGLES);
}
#endif
template <typename T> inline T clamp (T x, T a, T b)
{
return ((x) > (a) ? ((x) < (b) ? (x) : (b)) : (a));
}
template <typename T> inline T mapValue(T x, T a, T b, T c, T d)
{
x = clamp(x, a, b);
return c + (d - c) * (x - a) / (b - a);
}
static void getFlowField(const Mat& u, const Mat& v, Mat& flowField)
{
float maxDisplacement = 1.0f;
for (int i = 0; i < u.rows; ++i)
{
const float* ptr_u = u.ptr<float>(i);
const float* ptr_v = v.ptr<float>(i);
for (int j = 0; j < u.cols; ++j)
{
float d = max(fabsf(ptr_u[j]), fabsf(ptr_v[j]));
if (d > maxDisplacement)
maxDisplacement = d;
}
}
flowField.create(u.size(), CV_8UC4);
for (int i = 0; i < flowField.rows; ++i)
{
const float* ptr_u = u.ptr<float>(i);
const float* ptr_v = v.ptr<float>(i);
Vec4b* row = flowField.ptr<Vec4b>(i);
for (int j = 0; j < flowField.cols; ++j)
{
row[j][0] = 0;
row[j][1] = static_cast<unsigned char> (mapValue (-ptr_v[j], -maxDisplacement, maxDisplacement, 0.0f, 255.0f));
row[j][2] = static_cast<unsigned char> (mapValue ( ptr_u[j], -maxDisplacement, maxDisplacement, 0.0f, 255.0f));
row[j][3] = 255;
}
}
}
int main(int argc, const char* argv[])
{
const char* keys =
"{ h help | | print help message }"
"{ l left | | specify left image }"
"{ r right | | specify right image }"
"{ gray | | use grayscale sources [PyrLK Sparse] }"
"{ win_size | 21 | specify windows size [PyrLK] }"
"{ max_level | 3 | specify max level [PyrLK] }"
"{ iters | 30 | specify iterations count [PyrLK] }"
"{ points | 4000 | specify points count [GoodFeatureToTrack] }"
"{ min_dist | 0 | specify minimal distance between points [GoodFeatureToTrack] }";
CommandLineParser cmd(argc, argv, keys);
if (cmd.has("help") || !cmd.check())
{
cmd.printMessage();
cmd.printErrors();
return 0;
}
string fname0 = cmd.get<string>("left");
string fname1 = cmd.get<string>("right");
if (fname0.empty() || fname1.empty())
{
cerr << "Missing input file names" << endl;
return -1;
}
bool useGray = cmd.has("gray");
int winSize = cmd.get<int>("win_size");
int maxLevel = cmd.get<int>("max_level");
int iters = cmd.get<int>("iters");
int points = cmd.get<int>("points");
double minDist = cmd.get<double>("min_dist");
Mat frame0 = imread(fname0);
Mat frame1 = imread(fname1);
if (frame0.empty() || frame1.empty())
{
cout << "Can't load input images" << endl;
return -1;
}
namedWindow("PyrLK [Sparse]", WINDOW_NORMAL);
namedWindow("PyrLK [Dense] Flow Field", WINDOW_NORMAL);
#ifdef HAVE_OPENGL
namedWindow("PyrLK [Dense]", WINDOW_OPENGL);
setGlDevice();
#endif
cout << "Image size : " << frame0.cols << " x " << frame0.rows << endl;
cout << "Points count : " << points << endl;
cout << endl;
Mat frame0Gray;
cvtColor(frame0, frame0Gray, COLOR_BGR2GRAY);
Mat frame1Gray;
cvtColor(frame1, frame1Gray, COLOR_BGR2GRAY);
// goodFeaturesToTrack
GoodFeaturesToTrackDetector_GPU detector(points, 0.01, minDist);
GpuMat d_frame0Gray(frame0Gray);
GpuMat d_prevPts;
detector(d_frame0Gray, d_prevPts);
// Sparse
PyrLKOpticalFlow d_pyrLK;
d_pyrLK.winSize.width = winSize;
d_pyrLK.winSize.height = winSize;
d_pyrLK.maxLevel = maxLevel;
d_pyrLK.iters = iters;
GpuMat d_frame0(frame0);
GpuMat d_frame1(frame1);
GpuMat d_frame1Gray(frame1Gray);
GpuMat d_nextPts;
GpuMat d_status;
d_pyrLK.sparse(useGray ? d_frame0Gray : d_frame0, useGray ? d_frame1Gray : d_frame1, d_prevPts, d_nextPts, d_status);
// Draw arrows
vector<Point2f> prevPts(d_prevPts.cols);
download(d_prevPts, prevPts);
vector<Point2f> nextPts(d_nextPts.cols);
download(d_nextPts, nextPts);
vector<uchar> status(d_status.cols);
download(d_status, status);
drawArrows(frame0, prevPts, nextPts, status, Scalar(255, 0, 0));
imshow("PyrLK [Sparse]", frame0);
// Dense
GpuMat d_u;
GpuMat d_v;
d_pyrLK.dense(d_frame0Gray, d_frame1Gray, d_u, d_v);
// Draw flow field
Mat flowField;
getFlowField(Mat(d_u), Mat(d_v), flowField);
imshow("PyrLK [Dense] Flow Field", flowField);
#ifdef HAVE_OPENGL
setOpenGlContext("PyrLK [Dense]");
GpuMat d_vertex, d_colors;
createOpticalFlowNeedleMap(d_u, d_v, d_vertex, d_colors);
DrawData drawData;
drawData.tex.copyFrom(d_frame0Gray);
drawData.arr.setVertexArray(d_vertex);
drawData.arr.setColorArray(d_colors, false);
setOpenGlDrawCallback("PyrLK [Dense]", drawCallback, &drawData);
#endif
waitKey();
return 0;
}
#include <iostream>
#include <vector>
#include "cvconfig.h"
#include "opencv2/core/core.hpp"
#include "opencv2/core/opengl_interop.hpp"
#include "opencv2/imgproc/imgproc.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/video/video.hpp"
#include "opencv2/gpu/gpu.hpp"
using namespace std;
using namespace cv;
using namespace cv::gpu;
static void download(const GpuMat& d_mat, vector<Point2f>& vec)
{
vec.resize(d_mat.cols);
Mat mat(1, d_mat.cols, CV_32FC2, (void*)&vec[0]);
d_mat.download(mat);
}
static void download(const GpuMat& d_mat, vector<uchar>& vec)
{
vec.resize(d_mat.cols);
Mat mat(1, d_mat.cols, CV_8UC1, (void*)&vec[0]);
d_mat.download(mat);
}
static void drawArrows(Mat& frame, const vector<Point2f>& prevPts, const vector<Point2f>& nextPts, const vector<uchar>& status, Scalar line_color = Scalar(0, 0, 255))
{
for (size_t i = 0; i < prevPts.size(); ++i)
{
if (status[i])
{
int line_thickness = 1;
Point p = prevPts[i];
Point q = nextPts[i];
double angle = atan2((double) p.y - q.y, (double) p.x - q.x);
double hypotenuse = sqrt( (double)(p.y - q.y)*(p.y - q.y) + (double)(p.x - q.x)*(p.x - q.x) );
if (hypotenuse < 1.0)
continue;
// Here we lengthen the arrow by a factor of three.
q.x = (int) (p.x - 3 * hypotenuse * cos(angle));
q.y = (int) (p.y - 3 * hypotenuse * sin(angle));
// Now we draw the main line of the arrow.
line(frame, p, q, line_color, line_thickness);
// Now draw the tips of the arrow. I do some scaling so that the
// tips look proportional to the main line of the arrow.
p.x = (int) (q.x + 9 * cos(angle + CV_PI / 4));
p.y = (int) (q.y + 9 * sin(angle + CV_PI / 4));
line(frame, p, q, line_color, line_thickness);
p.x = (int) (q.x + 9 * cos(angle - CV_PI / 4));
p.y = (int) (q.y + 9 * sin(angle - CV_PI / 4));
line(frame, p, q, line_color, line_thickness);
}
}
}
#ifdef HAVE_OPENGL
struct DrawData
{
GlTexture tex;
GlArrays arr;
};
static void drawCallback(void* userdata)
{
DrawData* data = static_cast<DrawData*>(userdata);
if (data->tex.empty() || data->arr.empty())
return;
static GlCamera camera;
static bool init_camera = true;
if (init_camera)
{
camera.setOrthoProjection(0.0, 1.0, 1.0, 0.0, 0.0, 1.0);
camera.lookAt(Point3d(0.0, 0.0, 1.0), Point3d(0.0, 0.0, 0.0), Point3d(0.0, 1.0, 0.0));
init_camera = false;
}
camera.setupProjectionMatrix();
camera.setupModelViewMatrix();
render(data->tex);
render(data->arr, RenderMode::TRIANGLES);
}
#endif
template <typename T> inline T clamp (T x, T a, T b)
{
return ((x) > (a) ? ((x) < (b) ? (x) : (b)) : (a));
}
template <typename T> inline T mapValue(T x, T a, T b, T c, T d)
{
x = clamp(x, a, b);
return c + (d - c) * (x - a) / (b - a);
}
static void getFlowField(const Mat& u, const Mat& v, Mat& flowField)
{
float maxDisplacement = 1.0f;
for (int i = 0; i < u.rows; ++i)
{
const float* ptr_u = u.ptr<float>(i);
const float* ptr_v = v.ptr<float>(i);
for (int j = 0; j < u.cols; ++j)
{
float d = max(fabsf(ptr_u[j]), fabsf(ptr_v[j]));
if (d > maxDisplacement)
maxDisplacement = d;
}
}
flowField.create(u.size(), CV_8UC4);
for (int i = 0; i < flowField.rows; ++i)
{
const float* ptr_u = u.ptr<float>(i);
const float* ptr_v = v.ptr<float>(i);
Vec4b* row = flowField.ptr<Vec4b>(i);
for (int j = 0; j < flowField.cols; ++j)
{
row[j][0] = 0;
row[j][1] = static_cast<unsigned char> (mapValue (-ptr_v[j], -maxDisplacement, maxDisplacement, 0.0f, 255.0f));
row[j][2] = static_cast<unsigned char> (mapValue ( ptr_u[j], -maxDisplacement, maxDisplacement, 0.0f, 255.0f));
row[j][3] = 255;
}
}
}
int main(int argc, const char* argv[])
{
const char* keys =
"{ h help | | print help message }"
"{ l left | | specify left image }"
"{ r right | | specify right image }"
"{ gray | | use grayscale sources [PyrLK Sparse] }"
"{ win_size | 21 | specify windows size [PyrLK] }"
"{ max_level | 3 | specify max level [PyrLK] }"
"{ iters | 30 | specify iterations count [PyrLK] }"
"{ points | 4000 | specify points count [GoodFeatureToTrack] }"
"{ min_dist | 0 | specify minimal distance between points [GoodFeatureToTrack] }";
CommandLineParser cmd(argc, argv, keys);
if (cmd.has("help") || !cmd.check())
{
cmd.printMessage();
cmd.printErrors();
return 0;
}
string fname0 = cmd.get<string>("left");
string fname1 = cmd.get<string>("right");
if (fname0.empty() || fname1.empty())
{
cerr << "Missing input file names" << endl;
return -1;
}
bool useGray = cmd.has("gray");
int winSize = cmd.get<int>("win_size");
int maxLevel = cmd.get<int>("max_level");
int iters = cmd.get<int>("iters");
int points = cmd.get<int>("points");
double minDist = cmd.get<double>("min_dist");
Mat frame0 = imread(fname0);
Mat frame1 = imread(fname1);
if (frame0.empty() || frame1.empty())
{
cout << "Can't load input images" << endl;
return -1;
}
namedWindow("PyrLK [Sparse]", WINDOW_NORMAL);
namedWindow("PyrLK [Dense] Flow Field", WINDOW_NORMAL);
#ifdef HAVE_OPENGL
namedWindow("PyrLK [Dense]", WINDOW_OPENGL);
setGlDevice();
#endif
cout << "Image size : " << frame0.cols << " x " << frame0.rows << endl;
cout << "Points count : " << points << endl;
cout << endl;
Mat frame0Gray;
cvtColor(frame0, frame0Gray, COLOR_BGR2GRAY);
Mat frame1Gray;
cvtColor(frame1, frame1Gray, COLOR_BGR2GRAY);
// goodFeaturesToTrack
GoodFeaturesToTrackDetector_GPU detector(points, 0.01, minDist);
GpuMat d_frame0Gray(frame0Gray);
GpuMat d_prevPts;
detector(d_frame0Gray, d_prevPts);
// Sparse
PyrLKOpticalFlow d_pyrLK;
d_pyrLK.winSize.width = winSize;
d_pyrLK.winSize.height = winSize;
d_pyrLK.maxLevel = maxLevel;
d_pyrLK.iters = iters;
GpuMat d_frame0(frame0);
GpuMat d_frame1(frame1);
GpuMat d_frame1Gray(frame1Gray);
GpuMat d_nextPts;
GpuMat d_status;
d_pyrLK.sparse(useGray ? d_frame0Gray : d_frame0, useGray ? d_frame1Gray : d_frame1, d_prevPts, d_nextPts, d_status);
// Draw arrows
vector<Point2f> prevPts(d_prevPts.cols);
download(d_prevPts, prevPts);
vector<Point2f> nextPts(d_nextPts.cols);
download(d_nextPts, nextPts);
vector<uchar> status(d_status.cols);
download(d_status, status);
drawArrows(frame0, prevPts, nextPts, status, Scalar(255, 0, 0));
imshow("PyrLK [Sparse]", frame0);
// Dense
GpuMat d_u;
GpuMat d_v;
d_pyrLK.dense(d_frame0Gray, d_frame1Gray, d_u, d_v);
// Draw flow field
Mat flowField;
getFlowField(Mat(d_u), Mat(d_v), flowField);
imshow("PyrLK [Dense] Flow Field", flowField);
#ifdef HAVE_OPENGL
setOpenGlContext("PyrLK [Dense]");
GpuMat d_vertex, d_colors;
createOpticalFlowNeedleMap(d_u, d_v, d_vertex, d_colors);
DrawData drawData;
drawData.tex.copyFrom(d_frame0Gray);
drawData.arr.setVertexArray(d_vertex);
drawData.arr.setColorArray(d_colors, false);
setOpenGlDrawCallback("PyrLK [Dense]", drawCallback, &drawData);
#endif
waitKey();
return 0;
}

764
samples/gpu/stereo_match.cpp
View File

@@ -1,382 +1,382 @@
#include <iostream>
#include <string>
#include <sstream>
#include <iomanip>
#include <stdexcept>
#include "opencv2/gpu/gpu.hpp"
#include "opencv2/highgui/highgui.hpp"
using namespace cv;
using namespace std;
bool help_showed = false;
struct Params
{
Params();
static Params read(int argc, char** argv);
string left;
string right;
string method_str() const
{
switch (method)
{
case BM: return "BM";
case BP: return "BP";
case CSBP: return "CSBP";
}
return "";
}
enum {BM, BP, CSBP} method;
int ndisp; // Max disparity + 1
};
struct App
{
App(const Params& p);
void run();
void handleKey(char key);
void printParams() const;
void workBegin() { work_begin = getTickCount(); }
void workEnd()
{
int64 d = getTickCount() - work_begin;
double f = getTickFrequency();
work_fps = f / d;
}
string text() const
{
stringstream ss;
ss << "(" << p.method_str() << ") FPS: " << setiosflags(ios::left)
<< setprecision(4) << work_fps;
return ss.str();
}
private:
Params p;
bool running;
Mat left_src, right_src;
Mat left, right;
gpu::GpuMat d_left, d_right;
gpu::StereoBM_GPU bm;
gpu::StereoBeliefPropagation bp;
gpu::StereoConstantSpaceBP csbp;
int64 work_begin;
double work_fps;
};
static void printHelp()
{
cout << "Usage: stereo_match_gpu\n"
<< "\t--left <left_view> --right <right_view> # must be rectified\n"
<< "\t--method <stereo_match_method> # BM | BP | CSBP\n"
<< "\t--ndisp <number> # number of disparity levels\n";
help_showed = true;
}
int main(int argc, char** argv)
{
try
{
if (argc < 2)
{
printHelp();
return 1;
}
Params args = Params::read(argc, argv);
if (help_showed)
return -1;
App app(args);
app.run();
}
catch (const exception& e)
{
cout << "error: " << e.what() << endl;
}
return 0;
}
Params::Params()
{
method = BM;
ndisp = 64;
}
Params Params::read(int argc, char** argv)
{
Params p;
for (int i = 1; i < argc; i++)
{
if (string(argv[i]) == "--left") p.left = argv[++i];
else if (string(argv[i]) == "--right") p.right = argv[++i];
else if (string(argv[i]) == "--method")
{
if (string(argv[i + 1]) == "BM") p.method = BM;
else if (string(argv[i + 1]) == "BP") p.method = BP;
else if (string(argv[i + 1]) == "CSBP") p.method = CSBP;
else throw runtime_error("unknown stereo match method: " + string(argv[i + 1]));
i++;
}
else if (string(argv[i]) == "--ndisp") p.ndisp = atoi(argv[++i]);
else if (string(argv[i]) == "--help") printHelp();
else throw runtime_error("unknown key: " + string(argv[i]));
}
return p;
}
App::App(const Params& params)
: p(params), running(false)
{
cv::gpu::printShortCudaDeviceInfo(cv::gpu::getDevice());
cout << "stereo_match_gpu sample\n";
cout << "\nControls:\n"
<< "\tesc - exit\n"
<< "\tp - print current parameters\n"
<< "\tg - convert source images into gray\n"
<< "\tm - change stereo match method\n"
<< "\ts - change Sobel prefiltering flag (for BM only)\n"
<< "\t1/q - increase/decrease maximum disparity\n"
<< "\t2/w - increase/decrease window size (for BM only)\n"
<< "\t3/e - increase/decrease iteration count (for BP and CSBP only)\n"
<< "\t4/r - increase/decrease level count (for BP and CSBP only)\n";
}
void App::run()
{
// Load images
left_src = imread(p.left);
right_src = imread(p.right);
if (left_src.empty()) throw runtime_error("can't open file \"" + p.left + "\"");
if (right_src.empty()) throw runtime_error("can't open file \"" + p.right + "\"");
cvtColor(left_src, left, CV_BGR2GRAY);
cvtColor(right_src, right, CV_BGR2GRAY);
d_left.upload(left);
d_right.upload(right);
imshow("left", left);
imshow("right", right);
// Set common parameters
bm.ndisp = p.ndisp;
bp.ndisp = p.ndisp;
csbp.ndisp = p.ndisp;
// Prepare disparity map of specified type
Mat disp(left.size(), CV_8U);
gpu::GpuMat d_disp(left.size(), CV_8U);
cout << endl;
printParams();
running = true;
while (running)
{
workBegin();
switch (p.method)
{
case Params::BM:
if (d_left.channels() > 1 || d_right.channels() > 1)
{
cout << "BM doesn't support color images\n";
cvtColor(left_src, left, CV_BGR2GRAY);
cvtColor(right_src, right, CV_BGR2GRAY);
cout << "image_channels: " << left.channels() << endl;
d_left.upload(left);
d_right.upload(right);
imshow("left", left);
imshow("right", right);
}
bm(d_left, d_right, d_disp);
break;
case Params::BP: bp(d_left, d_right, d_disp); break;
case Params::CSBP: csbp(d_left, d_right, d_disp); break;
}
workEnd();
// Show results
d_disp.download(disp);
putText(disp, text(), Point(5, 25), FONT_HERSHEY_SIMPLEX, 1.0, Scalar::all(255));
imshow("disparity", disp);
handleKey((char)waitKey(3));
}
}
void App::printParams() const
{
cout << "--- Parameters ---\n";
cout << "image_size: (" << left.cols << ", " << left.rows << ")\n";
cout << "image_channels: " << left.channels() << endl;
cout << "method: " << p.method_str() << endl
<< "ndisp: " << p.ndisp << endl;
switch (p.method)
{
case Params::BM:
cout << "win_size: " << bm.winSize << endl;
cout << "prefilter_sobel: " << bm.preset << endl;
break;
case Params::BP:
cout << "iter_count: " << bp.iters << endl;
cout << "level_count: " << bp.levels << endl;
break;
case Params::CSBP:
cout << "iter_count: " << csbp.iters << endl;
cout << "level_count: " << csbp.levels << endl;
break;
}
cout << endl;
}
void App::handleKey(char key)
{
switch (key)
{
case 27:
running = false;
break;
case 'p': case 'P':
printParams();
break;
case 'g': case 'G':
if (left.channels() == 1 && p.method != Params::BM)
{
left = left_src;
right = right_src;
}
else
{
cvtColor(left_src, left, CV_BGR2GRAY);
cvtColor(right_src, right, CV_BGR2GRAY);
}
d_left.upload(left);
d_right.upload(right);
cout << "image_channels: " << left.channels() << endl;
imshow("left", left);
imshow("right", right);
break;
case 'm': case 'M':
switch (p.method)
{
case Params::BM:
p.method = Params::BP;
break;
case Params::BP:
p.method = Params::CSBP;
break;
case Params::CSBP:
p.method = Params::BM;
break;
}
cout << "method: " << p.method_str() << endl;
break;
case 's': case 'S':
if (p.method == Params::BM)
{
switch (bm.preset)
{
case gpu::StereoBM_GPU::BASIC_PRESET:
bm.preset = gpu::StereoBM_GPU::PREFILTER_XSOBEL;
break;
case gpu::StereoBM_GPU::PREFILTER_XSOBEL:
bm.preset = gpu::StereoBM_GPU::BASIC_PRESET;
break;
}
cout << "prefilter_sobel: " << bm.preset << endl;
}
break;
case '1':
p.ndisp = p.ndisp == 1 ? 8 : p.ndisp + 8;
cout << "ndisp: " << p.ndisp << endl;
bm.ndisp = p.ndisp;
bp.ndisp = p.ndisp;
csbp.ndisp = p.ndisp;
break;
case 'q': case 'Q':
p.ndisp = max(p.ndisp - 8, 1);
cout << "ndisp: " << p.ndisp << endl;
bm.ndisp = p.ndisp;
bp.ndisp = p.ndisp;
csbp.ndisp = p.ndisp;
break;
case '2':
if (p.method == Params::BM)
{
bm.winSize = min(bm.winSize + 1, 51);
cout << "win_size: " << bm.winSize << endl;
}
break;
case 'w': case 'W':
if (p.method == Params::BM)
{
bm.winSize = max(bm.winSize - 1, 2);
cout << "win_size: " << bm.winSize << endl;
}
break;
case '3':
if (p.method == Params::BP)
{
bp.iters += 1;
cout << "iter_count: " << bp.iters << endl;
}
else if (p.method == Params::CSBP)
{
csbp.iters += 1;
cout << "iter_count: " << csbp.iters << endl;
}
break;
case 'e': case 'E':
if (p.method == Params::BP)
{
bp.iters = max(bp.iters - 1, 1);
cout << "iter_count: " << bp.iters << endl;
}
else if (p.method == Params::CSBP)
{
csbp.iters = max(csbp.iters - 1, 1);
cout << "iter_count: " << csbp.iters << endl;
}
break;
case '4':
if (p.method == Params::BP)
{
bp.levels += 1;
cout << "level_count: " << bp.levels << endl;
}
else if (p.method == Params::CSBP)
{
csbp.levels += 1;
cout << "level_count: " << csbp.levels << endl;
}
break;
case 'r': case 'R':
if (p.method == Params::BP)
{
bp.levels = max(bp.levels - 1, 1);
cout << "level_count: " << bp.levels << endl;
}
else if (p.method == Params::CSBP)
{
csbp.levels = max(csbp.levels - 1, 1);
cout << "level_count: " << csbp.levels << endl;
}
break;
}
}
#include <iostream>
#include <string>
#include <sstream>
#include <iomanip>
#include <stdexcept>
#include "opencv2/gpu/gpu.hpp"
#include "opencv2/highgui/highgui.hpp"
using namespace cv;
using namespace std;
bool help_showed = false;
struct Params
{
Params();
static Params read(int argc, char** argv);
string left;
string right;
string method_str() const
{
switch (method)
{
case BM: return "BM";
case BP: return "BP";
case CSBP: return "CSBP";
}
return "";
}
enum {BM, BP, CSBP} method;
int ndisp; // Max disparity + 1
};
struct App
{
App(const Params& p);
void run();
void handleKey(char key);
void printParams() const;
void workBegin() { work_begin = getTickCount(); }
void workEnd()
{
int64 d = getTickCount() - work_begin;
double f = getTickFrequency();
work_fps = f / d;
}
string text() const
{
stringstream ss;
ss << "(" << p.method_str() << ") FPS: " << setiosflags(ios::left)
<< setprecision(4) << work_fps;
return ss.str();
}
private:
Params p;
bool running;
Mat left_src, right_src;
Mat left, right;
gpu::GpuMat d_left, d_right;
gpu::StereoBM_GPU bm;
gpu::StereoBeliefPropagation bp;
gpu::StereoConstantSpaceBP csbp;
int64 work_begin;
double work_fps;
};
static void printHelp()
{
cout << "Usage: stereo_match_gpu\n"
<< "\t--left <left_view> --right <right_view> # must be rectified\n"
<< "\t--method <stereo_match_method> # BM | BP | CSBP\n"
<< "\t--ndisp <number> # number of disparity levels\n";
help_showed = true;
}
int main(int argc, char** argv)
{
try
{
if (argc < 2)
{
printHelp();
return 1;
}
Params args = Params::read(argc, argv);
if (help_showed)
return -1;
App app(args);
app.run();
}
catch (const exception& e)
{
cout << "error: " << e.what() << endl;
}
return 0;
}
Params::Params()
{
method = BM;
ndisp = 64;
}
Params Params::read(int argc, char** argv)
{
Params p;
for (int i = 1; i < argc; i++)
{
if (string(argv[i]) == "--left") p.left = argv[++i];
else if (string(argv[i]) == "--right") p.right = argv[++i];
else if (string(argv[i]) == "--method")
{
if (string(argv[i + 1]) == "BM") p.method = BM;
else if (string(argv[i + 1]) == "BP") p.method = BP;
else if (string(argv[i + 1]) == "CSBP") p.method = CSBP;
else throw runtime_error("unknown stereo match method: " + string(argv[i + 1]));
i++;
}
else if (string(argv[i]) == "--ndisp") p.ndisp = atoi(argv[++i]);
else if (string(argv[i]) == "--help") printHelp();
else throw runtime_error("unknown key: " + string(argv[i]));
}
return p;
}
App::App(const Params& params)
: p(params), running(false)
{
cv::gpu::printShortCudaDeviceInfo(cv::gpu::getDevice());
cout << "stereo_match_gpu sample\n";
cout << "\nControls:\n"
<< "\tesc - exit\n"
<< "\tp - print current parameters\n"
<< "\tg - convert source images into gray\n"
<< "\tm - change stereo match method\n"
<< "\ts - change Sobel prefiltering flag (for BM only)\n"
<< "\t1/q - increase/decrease maximum disparity\n"
<< "\t2/w - increase/decrease window size (for BM only)\n"
<< "\t3/e - increase/decrease iteration count (for BP and CSBP only)\n"
<< "\t4/r - increase/decrease level count (for BP and CSBP only)\n";
}
void App::run()
{
// Load images
left_src = imread(p.left);
right_src = imread(p.right);
if (left_src.empty()) throw runtime_error("can't open file \"" + p.left + "\"");
if (right_src.empty()) throw runtime_error("can't open file \"" + p.right + "\"");
cvtColor(left_src, left, CV_BGR2GRAY);
cvtColor(right_src, right, CV_BGR2GRAY);
d_left.upload(left);
d_right.upload(right);
imshow("left", left);
imshow("right", right);
// Set common parameters
bm.ndisp = p.ndisp;
bp.ndisp = p.ndisp;
csbp.ndisp = p.ndisp;
// Prepare disparity map of specified type
Mat disp(left.size(), CV_8U);
gpu::GpuMat d_disp(left.size(), CV_8U);
cout << endl;
printParams();
running = true;
while (running)
{
workBegin();
switch (p.method)
{
case Params::BM:
if (d_left.channels() > 1 || d_right.channels() > 1)
{
cout << "BM doesn't support color images\n";
cvtColor(left_src, left, CV_BGR2GRAY);
cvtColor(right_src, right, CV_BGR2GRAY);
cout << "image_channels: " << left.channels() << endl;
d_left.upload(left);
d_right.upload(right);
imshow("left", left);
imshow("right", right);
}
bm(d_left, d_right, d_disp);
break;
case Params::BP: bp(d_left, d_right, d_disp); break;
case Params::CSBP: csbp(d_left, d_right, d_disp); break;
}
workEnd();
// Show results
d_disp.download(disp);
putText(disp, text(), Point(5, 25), FONT_HERSHEY_SIMPLEX, 1.0, Scalar::all(255));
imshow("disparity", disp);
handleKey((char)waitKey(3));
}
}
void App::printParams() const
{
cout << "--- Parameters ---\n";
cout << "image_size: (" << left.cols << ", " << left.rows << ")\n";
cout << "image_channels: " << left.channels() << endl;
cout << "method: " << p.method_str() << endl
<< "ndisp: " << p.ndisp << endl;
switch (p.method)
{
case Params::BM:
cout << "win_size: " << bm.winSize << endl;
cout << "prefilter_sobel: " << bm.preset << endl;
break;
case Params::BP:
cout << "iter_count: " << bp.iters << endl;
cout << "level_count: " << bp.levels << endl;
break;
case Params::CSBP:
cout << "iter_count: " << csbp.iters << endl;
cout << "level_count: " << csbp.levels << endl;
break;
}
cout << endl;
}
void App::handleKey(char key)
{
switch (key)
{
case 27:
running = false;
break;
case 'p': case 'P':
printParams();
break;
case 'g': case 'G':
if (left.channels() == 1 && p.method != Params::BM)
{
left = left_src;
right = right_src;
}
else
{
cvtColor(left_src, left, CV_BGR2GRAY);
cvtColor(right_src, right, CV_BGR2GRAY);
}
d_left.upload(left);
d_right.upload(right);
cout << "image_channels: " << left.channels() << endl;
imshow("left", left);
imshow("right", right);
break;
case 'm': case 'M':
switch (p.method)
{
case Params::BM:
p.method = Params::BP;
break;
case Params::BP:
p.method = Params::CSBP;
break;
case Params::CSBP:
p.method = Params::BM;
break;
}
cout << "method: " << p.method_str() << endl;
break;
case 's': case 'S':
if (p.method == Params::BM)
{
switch (bm.preset)
{
case gpu::StereoBM_GPU::BASIC_PRESET:
bm.preset = gpu::StereoBM_GPU::PREFILTER_XSOBEL;
break;
case gpu::StereoBM_GPU::PREFILTER_XSOBEL:
bm.preset = gpu::StereoBM_GPU::BASIC_PRESET;
break;
}
cout << "prefilter_sobel: " << bm.preset << endl;
}
break;
case '1':
p.ndisp = p.ndisp == 1 ? 8 : p.ndisp + 8;
cout << "ndisp: " << p.ndisp << endl;
bm.ndisp = p.ndisp;
bp.ndisp = p.ndisp;
csbp.ndisp = p.ndisp;
break;
case 'q': case 'Q':
p.ndisp = max(p.ndisp - 8, 1);
cout << "ndisp: " << p.ndisp << endl;
bm.ndisp = p.ndisp;
bp.ndisp = p.ndisp;
csbp.ndisp = p.ndisp;
break;
case '2':
if (p.method == Params::BM)
{
bm.winSize = min(bm.winSize + 1, 51);
cout << "win_size: " << bm.winSize << endl;
}
break;
case 'w': case 'W':
if (p.method == Params::BM)
{
bm.winSize = max(bm.winSize - 1, 2);
cout << "win_size: " << bm.winSize << endl;
}
break;
case '3':
if (p.method == Params::BP)
{
bp.iters += 1;
cout << "iter_count: " << bp.iters << endl;
}
else if (p.method == Params::CSBP)
{
csbp.iters += 1;
cout << "iter_count: " << csbp.iters << endl;
}
break;
case 'e': case 'E':
if (p.method == Params::BP)
{
bp.iters = max(bp.iters - 1, 1);
cout << "iter_count: " << bp.iters << endl;
}
else if (p.method == Params::CSBP)
{
csbp.iters = max(csbp.iters - 1, 1);
cout << "iter_count: " << csbp.iters << endl;
}
break;
case '4':
if (p.method == Params::BP)
{
bp.levels += 1;
cout << "level_count: " << bp.levels << endl;
}
else if (p.method == Params::CSBP)
{
csbp.levels += 1;
cout << "level_count: " << csbp.levels << endl;
}
break;
case 'r': case 'R':
if (p.method == Params::BP)
{
bp.levels = max(bp.levels - 1, 1);
cout << "level_count: " << bp.levels << endl;
}
else if (p.method == Params::CSBP)
{
csbp.levels = max(csbp.levels - 1, 1);
cout << "level_count: " << csbp.levels << endl;
}
break;
}
}

304
samples/gpu/stereo_multi.cpp
View File

@@ -1,152 +1,152 @@
/* This sample demonstrates working on one piece of data using two GPUs.
It splits input into two parts and processes them separately on different
GPUs. */
// Disable some warnings which are caused with CUDA headers
#if defined(_MSC_VER)
#pragma warning(disable: 4201 4408 4100)
#endif
#include <iostream>
#include "cvconfig.h"
#include "opencv2/core/core.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/gpu/gpu.hpp"
#if !defined(HAVE_CUDA) || !defined(HAVE_TBB)
int main()
{
#if !defined(HAVE_CUDA)
std::cout << "CUDA support is required (CMake key 'WITH_CUDA' must be true).\n";
#endif
#if !defined(HAVE_TBB)
std::cout << "TBB support is required (CMake key 'WITH_TBB' must be true).\n";
#endif
return 0;
}
#else
#include "opencv2/core/internal.hpp" // For TBB wrappers
using namespace std;
using namespace cv;
using namespace cv::gpu;
struct Worker { void operator()(int device_id) const; };
// GPUs data
GpuMat d_left[2];
GpuMat d_right[2];
StereoBM_GPU* bm[2];
GpuMat d_result[2];
// CPU result
Mat result;
void printHelp()
{
std::cout << "Usage: stereo_multi_gpu --left <image> --right <image>\n";
}
int main(int argc, char** argv)
{
if (argc < 5)
{
printHelp();
return -1;
}
int num_devices = getCudaEnabledDeviceCount();
if (num_devices < 2)
{
std::cout << "Two or more GPUs are required\n";
return -1;
}
for (int i = 0; i < num_devices; ++i)
{
cv::gpu::printShortCudaDeviceInfo(i);
DeviceInfo dev_info(i);
if (!dev_info.isCompatible())
{
std::cout << "GPU module isn't built for GPU #" << i << " ("
<< dev_info.name() << ", CC " << dev_info.majorVersion()
<< dev_info.minorVersion() << "\n";
return -1;
}
}
// Load input data
Mat left, right;
for (int i = 1; i < argc; ++i)
{
if (string(argv[i]) == "--left")
{
left = imread(argv[++i], CV_LOAD_IMAGE_GRAYSCALE);
CV_Assert(!left.empty());
}
else if (string(argv[i]) == "--right")
{
right = imread(argv[++i], CV_LOAD_IMAGE_GRAYSCALE);
CV_Assert(!right.empty());
}
else if (string(argv[i]) == "--help")
{
printHelp();
return -1;
}
}
// Split source images for processing on the GPU #0
setDevice(0);
d_left[0].upload(left.rowRange(0, left.rows / 2));
d_right[0].upload(right.rowRange(0, right.rows / 2));
bm[0] = new StereoBM_GPU();
// Split source images for processing on the GPU #1
setDevice(1);
d_left[1].upload(left.rowRange(left.rows / 2, left.rows));
d_right[1].upload(right.rowRange(right.rows / 2, right.rows));
bm[1] = new StereoBM_GPU();
// Execute calculation in two threads using two GPUs
int devices[] = {0, 1};
parallel_do(devices, devices + 2, Worker());
// Release the first GPU resources
setDevice(0);
imshow("GPU #0 result", Mat(d_result[0]));
d_left[0].release();
d_right[0].release();
d_result[0].release();
delete bm[0];
// Release the second GPU resources
setDevice(1);
imshow("GPU #1 result", Mat(d_result[1]));
d_left[1].release();
d_right[1].release();
d_result[1].release();
delete bm[1];
waitKey();
return 0;
}
void Worker::operator()(int device_id) const
{
setDevice(device_id);
bm[device_id]->operator()(d_left[device_id], d_right[device_id],
d_result[device_id]);
std::cout << "GPU #" << device_id << " (" << DeviceInfo().name()
<< "): finished\n";
}
#endif
/* This sample demonstrates working on one piece of data using two GPUs.
It splits input into two parts and processes them separately on different
GPUs. */
// Disable some warnings which are caused with CUDA headers
#if defined(_MSC_VER)
#pragma warning(disable: 4201 4408 4100)
#endif
#include <iostream>
#include "cvconfig.h"
#include "opencv2/core/core.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/gpu/gpu.hpp"
#if !defined(HAVE_CUDA) || !defined(HAVE_TBB)
int main()
{
#if !defined(HAVE_CUDA)
std::cout << "CUDA support is required (CMake key 'WITH_CUDA' must be true).\n";
#endif
#if !defined(HAVE_TBB)
std::cout << "TBB support is required (CMake key 'WITH_TBB' must be true).\n";
#endif
return 0;
}
#else
#include "opencv2/core/internal.hpp" // For TBB wrappers
using namespace std;
using namespace cv;
using namespace cv::gpu;
struct Worker { void operator()(int device_id) const; };
// GPUs data
GpuMat d_left[2];
GpuMat d_right[2];
StereoBM_GPU* bm[2];
GpuMat d_result[2];
// CPU result
Mat result;
void printHelp()
{
std::cout << "Usage: stereo_multi_gpu --left <image> --right <image>\n";
}
int main(int argc, char** argv)
{
if (argc < 5)
{
printHelp();
return -1;
}
int num_devices = getCudaEnabledDeviceCount();
if (num_devices < 2)
{
std::cout << "Two or more GPUs are required\n";
return -1;
}
for (int i = 0; i < num_devices; ++i)
{
cv::gpu::printShortCudaDeviceInfo(i);
DeviceInfo dev_info(i);
if (!dev_info.isCompatible())
{
std::cout << "GPU module isn't built for GPU #" << i << " ("
<< dev_info.name() << ", CC " << dev_info.majorVersion()
<< dev_info.minorVersion() << "\n";
return -1;
}
}
// Load input data
Mat left, right;
for (int i = 1; i < argc; ++i)
{
if (string(argv[i]) == "--left")
{
left = imread(argv[++i], CV_LOAD_IMAGE_GRAYSCALE);
CV_Assert(!left.empty());
}
else if (string(argv[i]) == "--right")
{
right = imread(argv[++i], CV_LOAD_IMAGE_GRAYSCALE);
CV_Assert(!right.empty());
}
else if (string(argv[i]) == "--help")
{
printHelp();
return -1;
}
}
// Split source images for processing on the GPU #0
setDevice(0);
d_left[0].upload(left.rowRange(0, left.rows / 2));
d_right[0].upload(right.rowRange(0, right.rows / 2));
bm[0] = new StereoBM_GPU();
// Split source images for processing on the GPU #1
setDevice(1);
d_left[1].upload(left.rowRange(left.rows / 2, left.rows));
d_right[1].upload(right.rowRange(right.rows / 2, right.rows));
bm[1] = new StereoBM_GPU();
// Execute calculation in two threads using two GPUs
int devices[] = {0, 1};
parallel_do(devices, devices + 2, Worker());
// Release the first GPU resources
setDevice(0);
imshow("GPU #0 result", Mat(d_result[0]));
d_left[0].release();
d_right[0].release();
d_result[0].release();
delete bm[0];
// Release the second GPU resources
setDevice(1);
imshow("GPU #1 result", Mat(d_result[1]));
d_left[1].release();
d_right[1].release();
d_result[1].release();
delete bm[1];
waitKey();
return 0;
}
void Worker::operator()(int device_id) const
{
setDevice(device_id);
bm[device_id]->operator()(d_left[device_id], d_right[device_id],
d_result[device_id]);
std::cout << "GPU #" << device_id << " (" << DeviceInfo().name()
<< "): finished\n";
}
#endif

166
samples/gpu/surf_keypoint_matcher.cpp
View File

@@ -1,83 +1,83 @@
#include <iostream>
#include "opencv2/core/core.hpp"
#include "opencv2/features2d/features2d.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/gpu/gpu.hpp"
using namespace std;
using namespace cv;
using namespace cv::gpu;
static void help()
{
cout << "\nThis program demonstrates using SURF_GPU features detector, descriptor extractor and BruteForceMatcher_GPU" << endl;
cout << "\nUsage:\n\tmatcher_simple_gpu --left <image1> --right <image2>" << endl;
}
int main(int argc, char* argv[])
{
if (argc != 5)
{
help();
return -1;
}
GpuMat img1, img2;
for (int i = 1; i < argc; ++i)
{
if (string(argv[i]) == "--left")
{
img1.upload(imread(argv[++i], CV_LOAD_IMAGE_GRAYSCALE));
CV_Assert(!img1.empty());
}
else if (string(argv[i]) == "--right")
{
img2.upload(imread(argv[++i], CV_LOAD_IMAGE_GRAYSCALE));
CV_Assert(!img2.empty());
}
else if (string(argv[i]) == "--help")
{
help();
return -1;
}
}
cv::gpu::printShortCudaDeviceInfo(cv::gpu::getDevice());
SURF_GPU surf;
// detecting keypoints & computing descriptors
GpuMat keypoints1GPU, keypoints2GPU;
GpuMat descriptors1GPU, descriptors2GPU;
surf(img1, GpuMat(), keypoints1GPU, descriptors1GPU);
surf(img2, GpuMat(), keypoints2GPU, descriptors2GPU);
cout << "FOUND " << keypoints1GPU.cols << " keypoints on first image" << endl;
cout << "FOUND " << keypoints2GPU.cols << " keypoints on second image" << endl;
// matching descriptors
BFMatcher_GPU matcher(NORM_L2);
GpuMat trainIdx, distance;
matcher.matchSingle(descriptors1GPU, descriptors2GPU, trainIdx, distance);
// downloading results
vector<KeyPoint> keypoints1, keypoints2;
vector<float> descriptors1, descriptors2;
vector<DMatch> matches;
surf.downloadKeypoints(keypoints1GPU, keypoints1);
surf.downloadKeypoints(keypoints2GPU, keypoints2);
surf.downloadDescriptors(descriptors1GPU, descriptors1);
surf.downloadDescriptors(descriptors2GPU, descriptors2);
BFMatcher_GPU::matchDownload(trainIdx, distance, matches);
// drawing the results
Mat img_matches;
drawMatches(Mat(img1), keypoints1, Mat(img2), keypoints2, matches, img_matches);
namedWindow("matches", 0);
imshow("matches", img_matches);
waitKey(0);
return 0;
}
#include <iostream>
#include "opencv2/core/core.hpp"
#include "opencv2/features2d/features2d.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/gpu/gpu.hpp"
using namespace std;
using namespace cv;
using namespace cv::gpu;
static void help()
{
cout << "\nThis program demonstrates using SURF_GPU features detector, descriptor extractor and BruteForceMatcher_GPU" << endl;
cout << "\nUsage:\n\tmatcher_simple_gpu --left <image1> --right <image2>" << endl;
}
int main(int argc, char* argv[])
{
if (argc != 5)
{
help();
return -1;
}
GpuMat img1, img2;
for (int i = 1; i < argc; ++i)
{
if (string(argv[i]) == "--left")
{
img1.upload(imread(argv[++i], CV_LOAD_IMAGE_GRAYSCALE));
CV_Assert(!img1.empty());
}
else if (string(argv[i]) == "--right")
{
img2.upload(imread(argv[++i], CV_LOAD_IMAGE_GRAYSCALE));
CV_Assert(!img2.empty());
}
else if (string(argv[i]) == "--help")
{
help();
return -1;
}
}
cv::gpu::printShortCudaDeviceInfo(cv::gpu::getDevice());
SURF_GPU surf;
// detecting keypoints & computing descriptors
GpuMat keypoints1GPU, keypoints2GPU;
GpuMat descriptors1GPU, descriptors2GPU;
surf(img1, GpuMat(), keypoints1GPU, descriptors1GPU);
surf(img2, GpuMat(), keypoints2GPU, descriptors2GPU);
cout << "FOUND " << keypoints1GPU.cols << " keypoints on first image" << endl;
cout << "FOUND " << keypoints2GPU.cols << " keypoints on second image" << endl;
// matching descriptors
BFMatcher_GPU matcher(NORM_L2);
GpuMat trainIdx, distance;
matcher.matchSingle(descriptors1GPU, descriptors2GPU, trainIdx, distance);
// downloading results
vector<KeyPoint> keypoints1, keypoints2;
vector<float> descriptors1, descriptors2;
vector<DMatch> matches;
surf.downloadKeypoints(keypoints1GPU, keypoints1);
surf.downloadKeypoints(keypoints2GPU, keypoints2);
surf.downloadDescriptors(descriptors1GPU, descriptors1);
surf.downloadDescriptors(descriptors2GPU, descriptors2);
BFMatcher_GPU::matchDownload(trainIdx, distance, matches);
// drawing the results
Mat img_matches;
drawMatches(Mat(img1), keypoints1, Mat(img2), keypoints2, matches, img_matches);
namedWindow("matches", 0);
imshow("matches", img_matches);
waitKey(0);
return 0;
}

142
samples/gpu/video_reader.cpp
View File

@@ -1,71 +1,71 @@
#include <iostream>
#include <string>
#include <vector>
#include <algorithm>
#include <numeric>
#include <opencv2/core/core.hpp>
#include <opencv2/core/opengl_interop.hpp>
#include <opencv2/gpu/gpu.hpp>
#include <opencv2/highgui/highgui.hpp>
#include <opencv2/contrib/contrib.hpp>
int main(int argc, const char* argv[])
{
if (argc != 2)
return -1;
const std::string fname(argv[1]);
cv::namedWindow("CPU", cv::WINDOW_NORMAL);
cv::namedWindow("GPU", cv::WINDOW_OPENGL);
cv::gpu::setGlDevice();
cv::Mat frame;
cv::VideoCapture reader(fname);
cv::gpu::GpuMat d_frame;
cv::gpu::VideoReader_GPU d_reader(fname);
d_reader.dumpFormat(std::cout);
cv::TickMeter tm;
std::vector<double> cpu_times;
std::vector<double> gpu_times;
for (;;)
{
tm.reset(); tm.start();
if (!reader.read(frame))
break;
tm.stop();
cpu_times.push_back(tm.getTimeMilli());
tm.reset(); tm.start();
if (!d_reader.read(d_frame))
break;
tm.stop();
gpu_times.push_back(tm.getTimeMilli());
cv::imshow("CPU", frame);
cv::imshow("GPU", d_frame);
if (cv::waitKey(3) > 0)
break;
}
if (!cpu_times.empty() && !gpu_times.empty())
{
std::cout << std::endl << "Results:" << std::endl;
std::sort(cpu_times.begin(), cpu_times.end());
std::sort(gpu_times.begin(), gpu_times.end());
double cpu_avg = std::accumulate(cpu_times.begin(), cpu_times.end(), 0.0) / cpu_times.size();
double gpu_avg = std::accumulate(gpu_times.begin(), gpu_times.end(), 0.0) / gpu_times.size();
std::cout << "CPU : Avg : " << cpu_avg << " ms FPS : " << 1000.0 / cpu_avg << std::endl;
std::cout << "GPU : Avg : " << gpu_avg << " ms FPS : " << 1000.0 / gpu_avg << std::endl;
}
return 0;
}
#include <iostream>
#include <string>
#include <vector>
#include <algorithm>
#include <numeric>
#include <opencv2/core/core.hpp>
#include <opencv2/core/opengl_interop.hpp>
#include <opencv2/gpu/gpu.hpp>
#include <opencv2/highgui/highgui.hpp>
#include <opencv2/contrib/contrib.hpp>
int main(int argc, const char* argv[])
{
if (argc != 2)
return -1;
const std::string fname(argv[1]);
cv::namedWindow("CPU", cv::WINDOW_NORMAL);
cv::namedWindow("GPU", cv::WINDOW_OPENGL);
cv::gpu::setGlDevice();
cv::Mat frame;
cv::VideoCapture reader(fname);
cv::gpu::GpuMat d_frame;
cv::gpu::VideoReader_GPU d_reader(fname);
d_reader.dumpFormat(std::cout);
cv::TickMeter tm;
std::vector<double> cpu_times;
std::vector<double> gpu_times;
for (;;)
{
tm.reset(); tm.start();
if (!reader.read(frame))
break;
tm.stop();
cpu_times.push_back(tm.getTimeMilli());
tm.reset(); tm.start();
if (!d_reader.read(d_frame))
break;
tm.stop();
gpu_times.push_back(tm.getTimeMilli());
cv::imshow("CPU", frame);
cv::imshow("GPU", d_frame);
if (cv::waitKey(3) > 0)
break;
}
if (!cpu_times.empty() && !gpu_times.empty())
{
std::cout << std::endl << "Results:" << std::endl;
std::sort(cpu_times.begin(), cpu_times.end());
std::sort(gpu_times.begin(), gpu_times.end());
double cpu_avg = std::accumulate(cpu_times.begin(), cpu_times.end(), 0.0) / cpu_times.size();
double gpu_avg = std::accumulate(gpu_times.begin(), gpu_times.end(), 0.0) / gpu_times.size();
std::cout << "CPU : Avg : " << cpu_avg << " ms FPS : " << 1000.0 / cpu_avg << std::endl;
std::cout << "GPU : Avg : " << gpu_avg << " ms FPS : " << 1000.0 / gpu_avg << std::endl;
}
return 0;
}

192
samples/gpu/video_writer.cpp
View File

@@ -1,96 +1,96 @@
#include <iostream>
#include <vector>
#include <numeric>
#include "opencv2/core/core.hpp"
#include "opencv2/gpu/gpu.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/contrib/contrib.hpp"
int main(int argc, const char* argv[])
{
if (argc != 2)
{
std::cerr << "Usage : video_writer <input video file>" << std::endl;
return -1;
}
const double FPS = 25.0;
cv::VideoCapture reader(argv[1]);
if (!reader.isOpened())
{
std::cerr << "Can't open input video file" << std::endl;
return -1;
}
cv::gpu::printShortCudaDeviceInfo(cv::gpu::getDevice());
cv::VideoWriter writer;
cv::gpu::VideoWriter_GPU d_writer;
cv::Mat frame;
cv::gpu::GpuMat d_frame;
std::vector<double> cpu_times;
std::vector<double> gpu_times;
cv::TickMeter tm;
for (int i = 1;; ++i)
{
std::cout << "Read " << i << " frame" << std::endl;
reader >> frame;
if (frame.empty())
{
std::cout << "Stop" << std::endl;
break;
}
if (!writer.isOpened())
{
std::cout << "Frame Size : " << frame.cols << "x" << frame.rows << std::endl;
std::cout << "Open CPU Writer" << std::endl;
if (!writer.open("output_cpu.avi", CV_FOURCC('X', 'V', 'I', 'D'), FPS, frame.size()))
return -1;
}
if (!d_writer.isOpened())
{
std::cout << "Open GPU Writer" << std::endl;
d_writer.open("output_gpu.avi", frame.size(), FPS);
}
d_frame.upload(frame);
std::cout << "Write " << i << " frame" << std::endl;
tm.reset(); tm.start();
writer.write(frame);
tm.stop();
cpu_times.push_back(tm.getTimeMilli());
tm.reset(); tm.start();
d_writer.write(d_frame);
tm.stop();
gpu_times.push_back(tm.getTimeMilli());
}
std::cout << std::endl << "Results:" << std::endl;
std::sort(cpu_times.begin(), cpu_times.end());
std::sort(gpu_times.begin(), gpu_times.end());
double cpu_avg = std::accumulate(cpu_times.begin(), cpu_times.end(), 0.0) / cpu_times.size();
double gpu_avg = std::accumulate(gpu_times.begin(), gpu_times.end(), 0.0) / gpu_times.size();
std::cout << "CPU [XVID] : Avg : " << cpu_avg << " ms FPS : " << 1000.0 / cpu_avg << std::endl;
std::cout << "GPU [H264] : Avg : " << gpu_avg << " ms FPS : " << 1000.0 / gpu_avg << std::endl;
return 0;
}
#include <iostream>
#include <vector>
#include <numeric>
#include "opencv2/core/core.hpp"
#include "opencv2/gpu/gpu.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/contrib/contrib.hpp"
int main(int argc, const char* argv[])
{
if (argc != 2)
{
std::cerr << "Usage : video_writer <input video file>" << std::endl;
return -1;
}
const double FPS = 25.0;
cv::VideoCapture reader(argv[1]);
if (!reader.isOpened())
{
std::cerr << "Can't open input video file" << std::endl;
return -1;
}
cv::gpu::printShortCudaDeviceInfo(cv::gpu::getDevice());
cv::VideoWriter writer;
cv::gpu::VideoWriter_GPU d_writer;
cv::Mat frame;
cv::gpu::GpuMat d_frame;
std::vector<double> cpu_times;
std::vector<double> gpu_times;
cv::TickMeter tm;
for (int i = 1;; ++i)
{
std::cout << "Read " << i << " frame" << std::endl;
reader >> frame;
if (frame.empty())
{
std::cout << "Stop" << std::endl;
break;
}
if (!writer.isOpened())
{
std::cout << "Frame Size : " << frame.cols << "x" << frame.rows << std::endl;
std::cout << "Open CPU Writer" << std::endl;
if (!writer.open("output_cpu.avi", CV_FOURCC('X', 'V', 'I', 'D'), FPS, frame.size()))
return -1;
}
if (!d_writer.isOpened())
{
std::cout << "Open GPU Writer" << std::endl;
d_writer.open("output_gpu.avi", frame.size(), FPS);
}
d_frame.upload(frame);
std::cout << "Write " << i << " frame" << std::endl;
tm.reset(); tm.start();
writer.write(frame);
tm.stop();
cpu_times.push_back(tm.getTimeMilli());
tm.reset(); tm.start();
d_writer.write(d_frame);
tm.stop();
gpu_times.push_back(tm.getTimeMilli());
}
std::cout << std::endl << "Results:" << std::endl;
std::sort(cpu_times.begin(), cpu_times.end());
std::sort(gpu_times.begin(), gpu_times.end());
double cpu_avg = std::accumulate(cpu_times.begin(), cpu_times.end(), 0.0) / cpu_times.size();
double gpu_avg = std::accumulate(gpu_times.begin(), gpu_times.end(), 0.0) / gpu_times.size();
std::cout << "CPU [XVID] : Avg : " << cpu_avg << " ms FPS : " << 1000.0 / cpu_avg << std::endl;
std::cout << "GPU [H264] : Avg : " << gpu_avg << " ms FPS : " << 1000.0 / gpu_avg << std::endl;
return 0;
}