You are lucky that I am currently teaching a class on this and just made a sample... but you will need to adapt it to your needs and make it more robust for your case

```

include <iostream>

include <opencv2 opencv.hpp="">

using namespace std; using namespace cv; using namespace cv::ml;

int main( int argc, const char** argv ) { /// Use the cmdlineparser to process input arguments CommandLineParser parser(argc, argv, "{ help h | | show this message }" "{ video v | | (required) path to video }" );

/// If help is entered
if (parser.has("help")){
    parser.printMessage();
    return 0;
}

/// Parse arguments
string video_location(parser.get<string>("video"));
if (video_location.empty()){
    parser.printMessage();
    return -1;
}

/// Create a videoreader interface
VideoCapture cap(video_location);
Mat current_frame;

/// Set up the pedestrian detector --> let us take the default one
HOGDescriptor hog;
hog.setSVMDetector(HOGDescriptor::getDefaultPeopleDetector());

/// Set up tracking vector
vector<Point> track;

while(true){
    /// Grab a single frame from the video
    cap >> current_frame;

    /// Check if the frame has content
    if(current_frame.empty()){
        cerr << "Video has ended or bad frame was read. Quitting." << endl;
        return 0;
    }

    /// run the detector with default parameters. to get a higher hit-rate
    /// (and more false alarms, respectively), decrease the hitThreshold and
    /// groupThreshold (set groupThreshold to 0 to turn off the grouping completely).

    ///image, vector of rectangles, hit threshold, win stride, padding, scale, group th
    Mat img = current_frame.clone();
    resize(img,img,Size(img.cols*2, img.rows*2));

    vector<Rect> found;
    vector<double> weights;

    hog.detectMultiScale(img, found, weights);

    /// draw detections and store location
    for( size_t i = 0; i < found.size(); i++ )
    {
        Rect r = found[i];
        rectangle(img, found[i], cv::Scalar(0,0,255), 3);
        stringstream temp;
        temp << weights[i];
        putText(img, temp.str(),Point(found[i].x,found[i].y+50), FONT_HERSHEY_SIMPLEX, 1, Scalar(0,0,255));
        track.push_back(Point(found[i].x+found[i].width/2,found[i].y+found[i].height/2));
    }

    /// plot the track so far
    for(size_t i = 1; i < track.size(); i++){
        line(img, track[i-1], track[i], Scalar(255,255,0), 2);
    }

    /// Show
    imshow("detected person", img);
    waitKey(1);
}

return 0;

}

```

You are lucky that I am currently teaching a class on this and just made a sample... but you will need to adapt it to your needs and make it more robust for your case

#include <iostream>
#include <opencv2/opencv.hpp>

using namespace std;
using namespace cv;
using namespace cv::ml;
 
cv::ml;

int main( int argc, const char** argv )
{
    /// Use the cmdlineparser to process input arguments
    CommandLineParser parser(argc, argv,
        "{ help h            |      | show this message }"
        "{ video v           |      | (required) path to video }"
    );
 
);

    /// If help is entered
 if (parser.has("help")){
     parser.printMessage();
     return 0;
 }

 /// Parse arguments
 string video_location(parser.get<string>("video"));
 if (video_location.empty()){
     parser.printMessage();
     return -1;
 }

 /// Create a videoreader interface
 VideoCapture cap(video_location);
 Mat current_frame;

 /// Set up the pedestrian detector --> let us take the default one
 HOGDescriptor hog;
 hog.setSVMDetector(HOGDescriptor::getDefaultPeopleDetector());

 /// Set up tracking vector
 vector<Point> track;

 while(true){
     /// Grab a single frame from the video
     cap >> current_frame;

     /// Check if the frame has content
     if(current_frame.empty()){
         cerr << "Video has ended or bad frame was read. Quitting." << endl;
         return 0;
     }

     /// run the detector with default parameters. to get a higher hit-rate
     /// (and more false alarms, respectively), decrease the hitThreshold and
     /// groupThreshold (set groupThreshold to 0 to turn off the grouping completely).

     ///image, vector of rectangles, hit threshold, win stride, padding, scale, group th
     Mat img = current_frame.clone();
     resize(img,img,Size(img.cols*2, img.rows*2));

     vector<Rect> found;
     vector<double> weights;

     hog.detectMultiScale(img, found, weights);

     /// draw detections and store location
     for( size_t i = 0; i < found.size(); i++ )
     {
         Rect r = found[i];
         rectangle(img, found[i], cv::Scalar(0,0,255), 3);
         stringstream temp;
         temp << weights[i];
         putText(img, temp.str(),Point(found[i].x,found[i].y+50), FONT_HERSHEY_SIMPLEX, 1, Scalar(0,0,255));
         track.push_back(Point(found[i].x+found[i].width/2,found[i].y+found[i].height/2));
     }

     /// plot the track so far
     for(size_t i = 1; i < track.size(); i++){
         line(img, track[i-1], track[i], Scalar(255,255,0), 2);
     }

     /// Show
     imshow("detected person", img);
     waitKey(1);
 }

 return 0;
}
  
}
 
```