Revision history [back]

I have just started with OpenCV. My motivation is object detection. I am using Bag of words algorithm http://docs.opencv.org/trunk/modules/features2d/doc/object_categorization.html

I am using caltech database. Partially Annotated Databases The Caltech Database http://pascallin.ecs.soton.ac.uk/challenges/VOC/databases.html

I have attached my working code. So far the recognition is based on MAT() which is formed after looping over all the positive dataset images. So everytime I introduce a new test image out of the training dataset the code build the new MAT(), bowTrainer.add(features), and new Vocabulary.

I want to remove this step of training again and again. So that when i give a new image it checks against "final trained matrix" and gives result quickly.

Also, how to implement it such a way that after predicting the class of input image system is trained afterwards??

#include "stdafx.h"
#include "opencv\cv.h" 
#include "opencv\highgui.h"
#include "opencv\ml.h"
#include <stdio.h>
#include <iostream>
#include <opencv2/features2d/features2d.hpp>
#include <opencv2/imgproc/imgproc.hpp>
#include <opencv2/highgui/highgui.hpp>
#include <opencv2\nonfree\nonfree.hpp>
#include <vector>

using namespace cv; 
using namespace std;

using std::cout;
using std::cerr;
using std::endl;
using std::vector;

char ch[30];

//--------Using SURF as feature extractor and FlannBased for assigning a new point to the nearest one in the dictionary
Ptr<DescriptorMatcher> matcher = DescriptorMatcher::create("FlannBased");
Ptr<DescriptorExtractor> extractor = new SurfDescriptorExtractor();
SurfFeatureDetector detector(500);

//---dictionary size=number of cluster's centroids
int dictionarySize = 1500;

TermCriteria tc(CV_TERMCRIT_ITER, 10, 0.001);
int retries = 1;
int flags = KMEANS_PP_CENTERS;
BOWKMeansTrainer bowTrainer(dictionarySize, tc, retries, flags);
BOWImgDescriptorExtractor bowDE(extractor, matcher);


// this function is being used to train the BOW classifier 
void collectclasscentroids() {
    IplImage *img;
    int i,j;
    for(j=1;j<=4;j++){
        for(i=1;i<=60;i++){
            sprintf( ch,"%s%d%s%d%s","train/",j," (",i,").jpg");
            const char* imageName = ch;
            img = cvLoadImage(imageName,0);
            vector<KeyPoint> keypoint;
            detector.detect(img, keypoint);
            Mat features;
            extractor->compute(img, keypoint, features);
            bowTrainer.add(features);
        }
    }
    return;
}



int _tmain(int argc, _TCHAR* argv[]){
    int i,j;
    IplImage *img2;
    cout<<"Vector quantization..."<<endl;

    collectclasscentroids();

    vector<Mat> descriptors = bowTrainer.getDescriptors();
    int count=0;
    for(vector<Mat>::iterator iter=descriptors.begin();iter!=descriptors.end();iter++){
        count+=iter->rows;
    }
    cout<<"Clustering "<<count<<" features"<<endl;

    //choosing cluster's centroids as dictionary's words
    Mat dictionary = bowTrainer.cluster();
    bowDE.setVocabulary(dictionary);

    cout<<"extracting histograms in the form of BOW for each image "<<endl;
    Mat labels(0, 1, CV_32FC1);
    Mat trainingData(0, dictionarySize, CV_32FC1);
    int k=0;
    vector<KeyPoint> keypoint1;
    Mat bowDescriptor1;

    //extracting histogram in the form of bow for each image 
    for(j=1;j<=4;j++){
        for(i=1;i<=60;i++){
            sprintf( ch,"%s%d%s%d%s","train/",j," (",i,").jpg");
            const char* imageName = ch;
            img2 = cvLoadImage(imageName,0);
            detector.detect(img2, keypoint1);
            bowDE.compute(img2, keypoint1, bowDescriptor1);
            trainingData.push_back(bowDescriptor1);
            //setting the label for images
            labels.push_back((float) j);
        }
    }

    //Setting up SVM parameters
    CvSVMParams params;
    params.kernel_type=CvSVM::RBF;
    params.svm_type=CvSVM::C_SVC;
    params.gamma=0.50625000000000009;
    params.C=312.50000000000000;
    params.term_crit=cvTermCriteria(CV_TERMCRIT_ITER,100,0.000001);
    CvSVM svm;
    cout << "Training SVM classifier\n";
    bool res = svm.train(trainingData,labels,cv::Mat(),cv::Mat(),params);

    cout<<"Processing evaluation data..."<<endl;
    Mat groundTruth(0, 1, CV_32FC1);
    Mat evalData(0, dictionarySize, CV_32FC1);
    k=0;
    vector<KeyPoint> keypoint2;
    Mat bowDescriptor2;
    Mat results(0, 1, CV_32FC1);;

    // Testing the training data
    /*for(j=1;j<=4;j++){
    for(i=1;i<=60;i++){
    sprintf( ch,"%s%d%s%d%s","eval/",j," (",i,").jpg");
    const char* imageName = ch;
    img2 = cvLoadImage(imageName,0);

    detector.detect(img2, keypoint2);
    bowDE.compute(img2, keypoint2, bowDescriptor2);

    evalData.push_back(bowDescriptor2);
    groundTruth.push_back((float) j);
    float response = svm.predict(bowDescriptor2);
    cout << response << endl;
    results.push_back(response);
    }
    }*/

    for(i=1;i<=4;i++){
        sprintf( ch,"%s%d%s","my_tests/",i,".jpg");
        const char* imageName = ch;
        img2 = cvLoadImage(imageName,0);

        detector.detect(img2, keypoint2);
        bowDE.compute(img2, keypoint2, bowDescriptor2);
        float response = svm.predict(bowDescriptor2);
        cout << response << endl;
    }
    cout <<"Hurrey" << endl;

    //calculate the number of unmatched classes 
    /*double errorRate = (double) countNonZero(groundTruth - results) / evalData.rows;
    printf("%s%f","Error rate is ",errorRate);*/

    return 0;

}

Thanks for your help. Aditi