bellonemauro / svm_pcl_demo Goto Github PK

Hi,
I want to train with my own data. Can I input the point cloud data (kitti) to train, or input the data for the extracted features, such as sift?

These are the errors I receive while trying to build (have you run into these before?):

Scanning dependencies of target pcl_SVM_train_classify
[ 50%] Building CXX object CMakeFiles/pcl_SVM_train_classify.dir/pcl_SVM_train_classify.cpp.o
/home/kdean/projects/src/SVM_pcl_demo/pcl_SVM_train_classify.cpp: In function ‘int main(int, char**)’:
/home/kdean/projects/src/SVM_pcl_demo/pcl_SVM_train_classify.cpp:117:45: error: ‘class pcl::SVMClassify’ has no member named ‘getParameters’
my_svm_parameters = my_svm_classifier.getParameters();
^
/home/kdean/projects/src/SVM_pcl_demo/pcl_SVM_train_classify.cpp:130:30: error: ‘class pcl::SVMClassify’ has no member named ‘hasLabelledTrainingSet’
if ( my_svm_classifier.hasLabelledTrainingSet())
^
/home/kdean/projects/src/SVM_pcl_demo/pcl_SVM_train_classify.cpp:176:7: error: ‘SVMtestReport’ is not a member of ‘pcl’
pcl::SVMtestReport svm_test_report = my_svm_classifier.getClassificationTestReport();
^
/home/kdean/projects/src/SVM_pcl_demo/pcl_SVM_train_classify.cpp:176:26: error: expected ‘;’ before ‘svm_test_report’
pcl::SVMtestReport svm_test_report = my_svm_classifier.getClassificationTestReport();
^
/home/kdean/projects/src/SVM_pcl_demo/pcl_SVM_train_classify.cpp:179:36: error: ‘svm_test_report’ was not declared in this scope
svm_test_report.accuracy,
^
/home/kdean/projects/src/SVM_pcl_demo/pcl_SVM_train_classify.cpp:211:23: error: ‘class pcl::SVMTrain’ has no member named ‘adaptProbToInput’
my_svm_trainer.adaptProbToInput();
^
/home/kdean/projects/src/SVM_pcl_demo/pcl_SVM_train_classify.cpp:267:67: error: could not convert ‘my_svm_trainer.pcl::SVMTrain::.pcl::SVM::saveClassifierModel(((const char*)"./model_out.dat"))’ from ‘void’ to ‘bool’
if ( my_svm_trainer.saveClassifierModel("./model_out.dat") ) {
^
/home/kdean/projects/src/SVM_pcl_demo/pcl_SVM_train_classify.cpp:352:31: error: ‘class pcl::SVMClassify’ has no member named ‘hasLabelledTrainingSet’
if ( my_svm_classifier.hasLabelledTrainingSet())
^
/home/kdean/projects/src/SVM_pcl_demo/pcl_SVM_train_classify.cpp:388:8: error: ‘SVMtestReport’ is not a member of ‘pcl’
pcl::SVMtestReport svm_test_report = my_svm_classifier.getClassificationTestReport();
^
/home/kdean/projects/src/SVM_pcl_demo/pcl_SVM_train_classify.cpp:388:27: error: expected ‘;’ before ‘svm_test_report’
pcl::SVMtestReport svm_test_report = my_svm_classifier.getClassificationTestReport();
^
/home/kdean/projects/src/SVM_pcl_demo/pcl_SVM_train_classify.cpp:391:36: error: ‘svm_test_report’ was not declared in this scope
svm_test_report.accuracy,
^
make[2]: *** [CMakeFiles/pcl_SVM_train_classify.dir/pcl_SVM_train_classify.cpp.o] Error 1
make[1]: *** [CMakeFiles/pcl_SVM_train_classify.dir/all] Error 2

Hi Mauro,

I didn't understand the labeling format, can you please explain the format?
how can we save tested results?

Thanks again

Hello ...
I currently want to perform SVM using 2d point cloud data.
I have acquired the data myself and created the dataset to match LibSVM, but it says that the input format is wrong.
Is this dataset wrong?
This is the first line of the dataset.

1 1:-0.893290 2:-0.109680 3:0.000000 4:-0.879160 5:-0.092400 6:0.000000 7:-0.874660 8:-0.076520 9:0.000000 10:-0.861900 11:-0.060270 12:0.000000 13:-0.861820 14:-0.045170 15:0.000000 16:-0.853480 17:-0.029800 18:0.000000 19:-0.846870 20:-0.014780 21:0.000000 22:-0.834000 23:-0.000000 24:0.000000 25:-0.827870 26:0.014450 27:0.000000 28:-0.814500 29:0.028440 30:0.000000 31:-0.810890 32:0.042500 33:0.000000 34:-0.803040 35:0.056150 36:0.000000 37:-0.799940 38:0.069990 39:0.000000 40:-0.811530 41:0.085300 42:0.000000 43:-0.833740 44:0.102370 45:0.000000 46:-0.838760 47:0.117880 48:0.000000 49:-0.860280 50:0.136250 51:0.000000 52:-0.868600 53:0.153160 54:0.000000 55:-0.880520 56:0.171160 57:0.000000 58:-0.884250 59:0.187950 60:0.000000 61:-0.883750 62:0.204030 63:0.000000 64:-0.901400 65:0.224750 66:0.000000

Hello,

I have some question about training my own model, could you give me some afvice

1.how can I create the test.dat and train.dat with my LiDAR?
2.which point cloud feature you use?

thank you so much

Hi Mauro,

Thanks for the code available for public.
I have a question regarding multiple classes, can we use it for classify the multiple classes.