blink.cpp

#include <opencv2/opencv.hpp>
#include <iostream>
#include <deque>
#include <map>
using namespace cv;
using namespace std;


int main(int argc, char **argv)
{
    string blink = "C:/data/blink/";
    string dummy;

    // let's use seperate label vectors, since your annotations have "closeRight" and "closeLeft"
    // (skipping those would mean, we're losing too many positives !)
    // initially assume
    // they're all "non-blink":
    vector<int> tlabels_L(15832,0);
    vector<int> tlabels_R(15832,0);
    // then fill in the ground-truth positives:
    ifstream anno(blink + "blink8_groundtruth.txt");
    //getline(anno, dummy);
    // ^ you do not need to skip the 1st line, imho.
  
    // this looks wrong, too:
    //while ( getline (anno,dummy) )
    
    while ( anno.good() )
    {
        int a;
        anno >> a;
        anno >> dummy ;
        if (dummy == "close")
        { // only "Full blink used here. data file contains frame, state (close)"
            tlabels_L[a] = 1;
            tlabels_R[a] = 1;
	}
        if (dummy == "closeLeft")
        {
	    tlabels_L[a] = 1;
	}
        if (dummy == "closeRight")
        { 
	    tlabels_R[a] = 1;
	}
    }

    // assuming, you have 2 txt files with your EAR values, 
    // let's read them, 
    // and write a 13-element time-series and label to csv.
    ofstream csv("ear.csv");
    ifstream in_L(blink + "EAR_OUTPUT_left_EYE.txt");
    ifstream in_R(blink + "EAR_OUTPUT_right_EYE.txt");
	
    deque<float> el,er; // our queues

	
    for (size_t i=0; i<15832; i++) 
    {
        if (! in_L.good()) 
	{
            cerr << "bad left file " << i << endl;
            break;
        }
        float ear_L;
	in_L >> ear_L;
	el.push_back(ear_L);

	if (! in_R.good()) 
	{
            cerr << "bad right file " << i << endl;
            break;
        }
        float ear_R;
	in_R >> ear_R;
	el.push_back(ear_R);
	    

        // write out queues for left & right eye
        // "backdated" to t-6 ! (center of the 13 elem queue)
        if (i >= 13) {
            csv << tlabels_L[i-6] << ",";
            for (int k=0; k<13; k++)
                csv << el[k] << ",";
            csv << endl;

            csv << tlabels_R[i-6] << ",";
            for (int k=0; k<13; k++)
                csv << er[k] << ",";
            csv << endl;

            el.pop_front();
            er.pop_front();
        }
    }

    // train & test with SVM:
    Ptr<ml::SVM> svm = ml::SVM::create();
    Ptr<ml::TrainData> tdata = ml::TrainData::loadFromCSV("ear.csv",0,0,1);
    tdata->setTrainTestSplitRatio(0.8);

    Mat data = tdata->getTrainSamples();
    Mat labels = tdata->getTrainResponses();
    labels.convertTo(labels, CV_32S); // hrrm!
    svm->train(data,0,labels);

    Mat vdata = tdata->getTestSamples();
    Mat vlabels = tdata->getTestResponses();
    Mat results;
    svm->predict(vdata,results);

    float correct = countNonZero(results == vlabels);
    float accuracy = correct / results.total();
    cerr << "accuracy: " << accuracy << endl;

    // accuracy alone is not enough here, since it might
    // simply have missed all positives !
    Mat_<float> confusion(2,2, 0.0f);
    for (int i=0; i<results.rows; i++) {
        int p = (int)results.at<float>(i);
        int t = (int)vlabels.at<float>(i);
        confusion(p,t) ++;
    }
    cerr << "confusion:\n" << confusion << endl;

    return 0;
}