Softkore's profile - activity

I am a fifth-year University art student and hobbyist programmer. I am fairly familiar with C++ and have managed to familiarize myself with OpenCV thanks to the abundance of fine resources. I am working on an ongoing project that will culminate in a museum installation.

So far I have created a program that tracks movement in live webcam video stream. The program can track many different objects by using absdiff comparisons on image frames, creating a threshold image, and analyzing the contours. I am using variables for threshold sensitivity and blur amount to somewhat control contour sizes. I have provided my source code at the bottom of this post.

The problem I have, is that a single person or object may be treated/tracked as many different objects based on lighting and other factors. I am interested in Cascade Classifier Training in order to identify the human figure. The background of my live stream is simple, monochromatic, static, and well-lit - so I don't imagine it will be hard to create negative training examples. The position of the camera will be unchanging, so I should be able to use it to take many still images of human figures throughout the space, for use as positive training examples.

I hope my understanding of Cascade Classifier Training is somewhat accurate. I am seeking simple beginner tutorials to instruct me step-by-step in training a classifier, with example code. Any help or resources that you can provide are of great help and interest, and I thank you in advance.

I am also interested in attempting to map the detected movement to a drawn space: although I am tracking the X & Y coordinates of the center of the contour objects, I have no way to, say, map the human figure's movement to a very basic rendered figure. This thread is likely not the place for this question, but perhaps someone looking at my cascade question can point me in the right direction.

Thing.h

#pragma once
#include <string>

using namespace std;

class Thing{

public:
    Thing();
    ~Thing();

    int getXPos();
    void setXPos(int x);

    int getYPos();
    void setYPos(int y);

private:
    int xPos, yPos;
    string type;
};

Thing.cpp

#include "Thing.h"

Thing::Thing(){}

Thing::~Thing(){}

int Thing::getXPos(){
    return Thing::xPos;
}

void Thing::setXPos(int x){
    Thing::xPos = x;
}

int Thing::getYPos(){
    return Thing::yPos;
}

void Thing::setYPos(int y){
    Thing::yPos = y;
}

main.cpp

#define _CRT_SECURE_NO_DEPRECATE

#include <sstream>
#include <string>
#include <iostream>
#include <vector>
#include <opencv\cv.h>
#include <opencv\highgui.h>
#include "Thing.h"

using namespace std;
using namespace cv;

const static int SENSITIVITY_VALUE = 10;
const static int BLUR_SIZE = 70;

const int FRAME_WIDTH = 640;
const int FRAME_HEIGHT = 480;

const int MIN_OBJECT_AREA = 20 * 20;


string intToString(int number){
    std::stringstream ss;
    ss << number;
    return ss.str();
}


void drawObject(vector<Thing> theThings, Mat &frame){

    for (int i = 0; i<theThings.size(); i++){
        cv::circle(frame, cv::Point(theThings.at(i).getXPos(), theThings.at(i).getYPos()), 10, cv::Scalar(0, 0, 255));
        cv::putText(frame, intToString(theThings ...