samples/tapi/hog.cpp

#include <iostream>
#include <fstream>
#include <string>
#include <sstream>
#include <iomanip>
#include <stdexcept>
#include <opencv2/core/ocl.hpp>
#include <opencv2/core/utility.hpp>
#include "opencv2/imgcodecs.hpp"
#include <opencv2/videoio.hpp>
#include <opencv2/highgui.hpp>
#include <opencv2/objdetect.hpp>
#include <opencv2/imgproc.hpp>
 
using namespace std;
using namespace cv;
 
class App
{
public:
App(CommandLineParser& cmd);
    void run();
    void handleKey(char key);
    void hogWorkBegin();
    void hogWorkEnd();
    string hogWorkFps() const;
    void workBegin();
    void workEnd();
    string workFps() const;
private:
App operator=(App&);
 
    //Args args;
    bool running;
    bool make_gray;
    double scale;
    double resize_scale;
    int win_width;
    int win_stride_width, win_stride_height;
    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;
 
    string img_source;
    string vdo_source;
    string output;
    int camera_id;
    bool write_once;
};
 
int main(int argc, char** argv)
{
    const char* keys =
        "{ h help | | 打印帮助信息 }"
        "{ i input | | 指定输入图像}"
        "{ c camera | -1 | 启用摄像头捕获 }"
        "{ v video | vtest.avi | 使用视频作为输入 }"
        "{ g gray | | 是否将图像转换为灰度图像}"
        "{ s scale | 1.0 | 检测前调整图像大小}"
        "{ o output | output.avi | 指定输入为图像时的输出路径}";
    CommandLineParser cmd(argc, argv, keys);
    if (cmd.has("help"))
    {
cmd.printMessage();
        return EXIT_SUCCESS;
    }
 
App app(cmd);
    try
    {
app.run();
    }
    catch (const Exception& e)
    {
        return cout << "错误: " << e.what() << endl, 1;
    }
    catch (const exception& e)
    {
        return cout << "错误: " << e.what() << endl, 1;
    }
    catch(...)
    {
        return cout << "未知异常" << endl, 1;
    }
    return EXIT_SUCCESS;
}
 
App::App(CommandLineParser& cmd)
{
cout << "\n控制键:\n"
<< "\tESC - 退出\n"
<< "\tm - 切换模式 GPU <-> CPU\n"
<< "\tg - 是否将图像转换为灰度图像\n"
<< "\to - 单次保存输出图像，或切换视频保存开关\n"
<< "\t1/q - 增加/减少 HOG 比例\n"
<< "\t2/w - 增加/减少层数\n"
<< "\t3/e - 增加/减少 HOG 组阈值\n"
<< "\t4/r - 增加/减少命中阈值\n"
<< endl;
 
make_gray = cmd.has("gray");
resize_scale = cmd.get<double>("s");
vdo_source = samples::findFileOrKeep(cmd.get<string>("v"));
img_source = cmd.get<string>("i");
output = cmd.get<string>("o");
camera_id = cmd.get<int>("c");
 
win_width = 48;
win_stride_width = 8;
win_stride_height = 8;
gr_threshold = 8;
nlevels = 13;
hit_threshold = 1.4;
    scale = 1.05;
gamma_corr = true;
write_once = false;
 
cout << "组阈值: " << gr_threshold << endl;
cout << "层数: " << nlevels << endl;
cout << "窗口宽度: " << win_width << endl;
cout << "窗口步长: (" << win_stride_width << ", " << win_stride_height << ")\n";
cout << "命中阈值: " << hit_threshold << endl;
cout << "伽马校正: " << gamma_corr << endl;
cout << endl;
}
 
void App::run()
{
running = true;
    VideoWriter video_writer;
 
    Size win_size(win_width, win_width * 2);
    Size win_stride(win_stride_width, win_stride_height);
 
    // 在此处创建 HOG 描述符和检测器
 
    HOGDescriptor 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);
hog.setSVMDetector( HOGDescriptor::getDaimlerPeopleDetector() );
 
    当 running 为真时
    {
        VideoCapture vc;
        UMat frame;
 
        如果 vdo_source 不为空字符串
        {
vc.open(vdo_source.c_str());
            如果 !vc.isOpened()
                抛出 runtime_error 异常 ("无法打开视频文件: " + vdo_source);
vc >> frame;
        }
        否则如果 camera_id 不等于 -1
        {
vc.open(camera_id);
            如果 !vc.isOpened()
            {
stringstream msg;
msg << "无法打开摄像头: " << camera_id;
                抛出 runtime_error 异常(msg.str());
            }
vc >> frame;
        }
        否则
        {
            imread(img_source).copyTo(frame);
            如果 frame.empty()
                抛出 runtime_error 异常 ("无法打开图像文件: " + img_source);
        }
 
        UMat img_aux, img, img_to_show;
 
        // 迭代所有帧
        当 running 为真且 !frame.empty() 时
        {
workBegin();
 
            // 改变图像格式
            如果 make_gray 为真 cvtColor(frame, img_aux, COLOR_BGR2GRAY );
            否则 frame.copyTo(img_aux);
 
            // 调整图像大小
            如果 abs(scale-1.0)>0.001
            {
                Size sz((int)((double)img_aux.cols/resize_scale), (int)((double)img_aux.rows/resize_scale));
                resize(img_aux, img, sz, 0, 0, INTER_LINEAR_EXACT);
            }
            否则 img = img_aux;
img.copyTo(img_to_show);
hog.nlevels = nlevels;
vector<Rect> found;
 
            // 执行 HOG 分类
hogWorkBegin();
 
hog.detectMultiScale(img, found, hit_threshold, win_stride,
                    Size(0, 0), scale, gr_threshold);
hogWorkEnd();
 
 
            // 绘制正分类窗口
            对于 size_t i = 0; i < found.size(); i++
            {
                rectangle(img_to_show, found[i], Scalar(0, 255, 0), 3);
            }
 
            putText(img_to_show, ocl::useOpenCL() ? "模式: OpenCL" : "模式: CPU", Point(5, 25), FONT_HERSHEY_SIMPLEX, 1., Scalar(255, 100, 0), 2);
            putText(img_to_show, "FPS (仅 HOG): " + hogWorkFps(), Point(5, 65), FONT_HERSHEY_SIMPLEX, 1., Scalar(255, 100, 0), 2);
            putText(img_to_show, "FPS (总计): " + workFps(), Point(5, 105), FONT_HERSHEY_SIMPLEX, 1., Scalar(255, 100, 0), 2);
            imshow("opencv_hog", img_to_show);
            如果 vdo_source 不为空字符串 或 camera_id 不等于 -1  vc >> frame;
 
workEnd();
 
            如果 output 不为空字符串 且 write_once 为真
            {
                如果 img_source 不为空字符串 // 写入图像
                {
write_once = false;
                    imwrite(output, img_to_show);
                }
                否则 // 写入视频
                {
                    如果 !video_writer.isOpened()
                    {
video_writer.open(output, VideoWriter::fourcc('x','v','i','d'), 24,
img_to_show.size(), true);
                        如果 !video_writer.isOpened()
                            抛出 std::runtime_error 异常("无法创建视频写入器");
                    }
 
                    如果 make_gray 为真 cvtColor(img_to_show, img, COLOR_GRAY2BGR);
                    否则 cvtColor(img_to_show, img, COLOR_BGRA2BGR);
 
video_writer << img;
                }
            }
 
handleKey((char)waitKey(3));
        }
    }
}
 
void App::handleKey(char key)
{
    根据 key 的值执行以下操作
    {
    情况 27:
running = false;
        break;
    情况 'm'
    情况 'M'
ocl::setUseOpenCL(!cv::ocl::useOpenCL());
cout << "已切换到 " << (ocl::useOpenCL() ? "OpenCL 已启用" : "CPU") << " 模式\n";
        break;
    case 'g'
    case 'G'
make_gray = !make_gray;
cout << "将图像转换为灰度: " << (make_gray ? "YES" : "NO") << endl;
        break;
    case '1'
        scale *= 1.05;
cout << "缩放比例: " << scale << endl;
        break;
    case 'q'
    case 'Q'
        scale /= 1.05;
cout << "缩放比例: " << scale << endl;
        break;
    case '2'
nlevels++;
cout << "层数: " << nlevels << endl;
        break;
    case 'w'
    case 'W'
nlevels = max(nlevels - 1, 1);
cout << "层数: " << nlevels << endl;
        break;
    case '3'
gr_threshold++;
cout << "组阈值: " << gr_threshold << endl;
        break;
    case 'e'
    case 'E'
gr_threshold = max(0, gr_threshold - 1);
cout << "组阈值: " << gr_threshold << endl;
        break;
    case '4'
hit_threshold+=0.25;
cout << "命中阈值: " << hit_threshold << endl;
        break;
    case 'r'
    case 'R'
hit_threshold = max(0.0, hit_threshold - 0.25);
cout << "命中阈值: " << hit_threshold << endl;
        break;
    case 'c'
    case 'C'
gamma_corr = !gamma_corr;
cout << "伽马校正: " << gamma_corr << endl;
        break;
    case 'o'
    case 'O'
write_once = !write_once;
        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();
}