error in train casacde

edit

I did some background checking of the current OpenCV2.4.6.1 source code. Your problem is exactly that he cannot read the negative files, so there is actually something wrong with your bg.txt file.

Looking at cascadeclassifier.cpp we see this code:

for( int i = startNumStages; i < numStages; i++ )
    {
        cout << endl << "===== TRAINING " << i << "-stage =====" << endl;
        cout << "<BEGIN" << endl;

        if ( !updateTrainingSet( tempLeafFARate ) )
        {
        cout << "Train dataset for temp stage can not be filled. "
            "Branch training terminated." << endl;
        break;
    } 
    ...

This says clearly that the problem must be at the updateTrainingSet function for the first iteration. If we dig deeper, we can see this code:

bool CvCascadeClassifier::updateTrainingSet( double& acceptanceRatio)
{
    int64 posConsumed = 0, negConsumed = 0;
    imgReader.restart();
    int posCount = fillPassedSamples( 0, numPos, true, posConsumed );
    if( !posCount )
        return false;
    cout << "POS count : consumed   " << posCount << " : " << (int)posConsumed << endl;

    int proNumNeg = cvRound( ( ((double)numNeg) * ((double)posCount) ) / numPos ); // apply only a fraction of negative samples. double is required since overflow is possible
    int negCount = fillPassedSamples( posCount, proNumNeg, false, negConsumed );
    if ( !negCount )
        return false;

    curNumSamples = posCount + negCount;
    acceptanceRatio = negConsumed == 0 ? 0 : ( (double)negCount/(double)(int64)negConsumed );
    cout << "NEG count : acceptanceRatio    " << negCount << " : " << acceptanceRatio << endl;
    return true;
}

If you look at that code and your output, you clearly see that the positive samples are fetched and this works. The only other way of returning a false in this function, is if negCount has no value, meaning that the fillPassedSamples function doesn't work on your data :)

So please put your bg.txt file and all your path related info inside your question.

Ok so new answer. First of all, you are still using OpenCV2.4.3 since that is the dll type the exe requires when I am starting up your training system. I have switched to OpenCV2.4.6 (do this also) and compiled against the newest libraries, used the training algorithm on your data and it works perfect.

Be sure that your result folder is manually created before running the algorithm, else it cannot write data to it.

Here you can find the command I used to start the training, and the actual result of the first 2 and a half stages.

Don't mind the fact that you will not get the calculation time for the previous stage, it is something that I adapted in the source code, in order to see how my progress was going.

I found it has some empty lines in the end of your infofile.txt. It maybe the answer.

Hi, I had this similar problem which gives error:

POS count : consumed   50 : 50
Train dataset for temp stage can not be filled. Branch training terminated.

The problem was that, my bg.txt was generated in a Windows system using \r for next line. When I tried to use opencv_traincascade in Ubuntu, it read '\r' into the string for filelist, thus in CvCascadeImageReader::NegReader::nextImg() , the line src = imread(imgFilenames[last++], 0); did not work.

My fix was to add str.erase(std::remove(str.begin(), str.end(), '\r'), str.end()); before imgFilenames.push_back(dirname + str); in imagestorage.cpp

I hope this helps.

Hi Pinky, I met the same problem with you, how would you solve the problem? If the solution of it, can you tell me the solution, thank you!