floyd's profile - activity

i'm trying to do a haartraining using the following command:

opencv_traincascade -data xml -vec vec.vec -bg negext3000.txt -numPos 500 -numNeg 3000 -precalcValBufSize 1024 -precalcIdxBufSize 2048 -featureType LBP -w 48 -h 48

but always it crashes and i have to break it manually or it is runs endless and nothing more is happen.

PARAMETERS:
cascadeDirName: xml
vecFileName: vec.vec
bgFileName: negext3000.txt
numPos: 500
numNeg: 3000
numStages: 20
precalcValBufSize[Mb] : 1024
precalcIdxBufSize[Mb] : 2048
acceptanceRatioBreakValue : -1
stageType: BOOST
featureType: LBP
sampleWidth: 48
sampleHeight: 48
boostType: GAB
minHitRate: 0.995
maxFalseAlarmRate: 0.5
weightTrimRate: 0.95
maxDepth: 1
maxWeakCount: 100

===== TRAINING 0-stage =====
<BEGIN
POS count : consumed   500 : 500
NEG count : acceptanceRatio    3000 : 1
^C

I have tried so much:

• various values for -w -h
• different background images and sizes
• different num of stages
• different vec-files (with background, without background, with distortion)
• different num values for the vec-files
• …

i use a OS X 10.7.5 and open cv 3.1.0 so what can i try to get it running on OSX? should i switch to open cv 2.4.12?

in which direction i have to think to solve that problem? with which functions is that may possible? what else do i may need? it is not about face detection. the objects are from the same and if that is solved from different classes.

thanks in advance for any suggestions.

floyd

indeed i did have a unused map of svm's left in the code. all problems are gone. a new problem appeared, but so far for today and the new problem i can handle on my own. thanks a lot. :D :D

that's the problem, i do not know what is going on. the only error appearing is the title of that question. the location of the error is in the header file stl_pair.h. That is why i wrote my code seem to be okay. the location is out of my code. and all the informations shown in the screenshot are out of my code, too.

I'm doing like you did it here: link, a pair i do nowhere have.

i do agree with you my code is probably (or definitely) not okay. but way is there no reference to my code? so i don't know which part of the code i need to post.

hey everybody,

i am really new to openCV and i am trying to implement a surf detector using openCV 3.1 and i getting always after building my project in eclipse the error message above. There are a lot of information i getting, too. Most of them concerning to ml.hpp.

My code seem to be okay. The only error appearing references to "line 102, external location: /usr/local/Cellar/gcc48/4.8.2/lib/gcc/x86_64-apple-darwin11.4.2/4.8.2/include/c++/bits/stl_pair.h C/C++ Problem"

I do not know what to do cause the error is not my fault?

desperate floyd

EDIT 1

void categorizer::train_classifiers() {
// Set the vocabulary for the BOW descriptor extractor
bowDescriptorExtractor -> setVocabulary(vocab);
// Extract BOW descriptors for all training images and organize them into positive and negative samples for each category
make_pos_neg();

for(int i = 0; i < categories; i++) {
    string category = category_names[i];

    // Positive training data has labels 1
    Mat train_data = positive_data[category];
    Mat train_labels = Mat::ones(train_data.rows, 1, CV_32S);
    // Negative training data has labels 0
    train_data.push_back(negative_data[category]);
    Mat m = Mat::zeros(negative_data[category].rows, 1, CV_32S);
    train_labels.push_back(m);

    svm = ml::SVM::create();
    svm->setType(ml::SVM::C_SVC);
    svm->setKernel(ml::SVM::POLY);
    svm->setGamma(3);

    svm->train( train_data , ml::ROW_SAMPLE , train_labels );

    cout << "Trained and saved SVM for category " << category << endl;
}

}