Ask Your Question

Revision history [back]

click to hide/show revision 1
initial version

coin date cascade training

I have been trying to create a cascade classifier for a penny date prefix of the number 19. I am following this tutoria: http://coding-robin.de/2013/07/22/train-your-own-opencv-haar-classifier.html , but but I haven't had a positive result from detection. I am getting a circle in the middle regardless of what image i use.

C:\fakepath\opencv_penny_date_facedetect.png

Can someone please see if I am missing something?

Here are the files used for the training:

http://coinmatico.com/wp-content/uploads/2014/05/opencv-haar-classifier-training.zip

created samples like this:

perl bin/createsamples.pl positives.txt negatives.txt samples 1500\
  "opencv_createsamples -bgcolor 0 -bgthresh 0 -maxxangle 1.1\
  -maxyangle 1.1 maxzangle 0.5 -maxidev 40 -w 15 -h 15"

and trained like this:

opencv_traincascade -data classifier -vec samples.vec -bg negatives.txt\
  -numStages 20 -minHitRate 0.999 -maxFalseAlarmRate 0.5 -numPos 100\
  -numNeg 12 -w 15 -h 15 -mode ALL -precalcValBufSize 1024\
  -precalcIdxBufSize 1024

The classifier/cascade.xml is what i'm using with the facedetect.cpp sample from opencv: https://github.com/Itseez/opencv/blob/master/samples/c/facedetect.cpp

...I know I need a lot more sample images but I wanted to test something simple before I comited to all the cropping needed. Thanks for your help.

coin date cascade training

I have been trying to create a cascade classifier for a penny date prefix of the number 19. I am following this tutoria: http://coding-robin.de/2013/07/22/train-your-own-opencv-haar-classifier.html , but but I haven't had a positive result from detection. I am getting a circle in the middle regardless of what image i use.

C:\fakepath\opencv_penny_date_facedetect.png

Can someone please see if I am missing something?

Here are the files used for the training:

http://coinmatico.com/wp-content/uploads/2014/05/opencv-haar-classifier-training.zip

created samples like this:

perl bin/createsamples.pl positives.txt negatives.txt samples 1500\ "opencv_createsamples -bgcolor 0 -bgthresh 0 -maxxangle 1.1\ -maxyangle 1.1 maxzangle 0.5 -maxidev 40 -w 15 -h 15"

15"

and trained like this:

opencv_traincascade -data classifier -vec samples.vec -bg negatives.txt\ -numStages 20 -minHitRate 0.999 -maxFalseAlarmRate 0.5 -numPos 100\ -numNeg 12 -w 15 -h 15 -mode ALL -precalcValBufSize 1024\ -precalcIdxBufSize 1024

1024

The classifier/cascade.xml is what i'm using with the facedetect.cpp sample from opencv: https://github.com/Itseez/opencv/blob/master/samples/c/facedetect.cpp

...I know I need a lot more sample images but I wanted to test something simple before I comited to all the cropping needed. Thanks for your help.

coin date cascade training

I have been trying to create a cascade classifier for a penny date prefix of the number 19. I am following this tutoria: http://coding-robin.de/2013/07/22/train-your-own-opencv-haar-classifier.html , but but I haven't had a positive result from detection. I am getting a circle in the middle regardless of what image i use.

C:\fakepath\opencv_penny_date_facedetect.png

Can someone please see if I am missing something?

Here are the files used for the training:

http://coinmatico.com/wp-content/uploads/2014/05/opencv-haar-classifier-training.zip

created samples like this:

perl bin/createsamples.pl positives.txt negatives.txt samples 1500\
  "opencv_createsamples -bgcolor 0 -bgthresh 0 -maxxangle 1.1\
  -maxyangle 1.1 maxzangle 0.5 -maxidev 40 -w 15 -h 15"

15"

and trained like this:

opencv_traincascade -data classifier -vec samples.vec -bg negatives.txt\
  -numStages 20 -minHitRate 0.999 -maxFalseAlarmRate 0.5 -numPos 100\
  -numNeg 12 -w 15 -h 15 -mode ALL -precalcValBufSize 1024\
  -precalcIdxBufSize 1024

1024

The classifier/cascade.xml is what i'm using with the facedetect.cpp sample from opencv: https://github.com/Itseez/opencv/blob/master/samples/c/facedetect.cpp

...I know I need a lot more sample images but I wanted to test something simple before I comited to all the cropping needed. Thanks for your help.

coin date cascade training

I have been trying to create a cascade classifier for a penny date prefix of the number 19. I am following this tutoria: http://coding-robin.de/2013/07/22/train-your-own-opencv-haar-classifier.html , but but I haven't had a positive result from detection. I am getting a circle in the middle regardless of what image i use.

C:\fakepath\opencv_penny_date_facedetect.png

Can someone please see if I am missing something?

Here are the files used for the training:

http://coinmatico.com/wp-content/uploads/2014/05/opencv-haar-classifier-training.ziphttp://coinmatico.com/wp-content/uploads/2014/05/opencv-haar-classifier-training2.zip

created samples like this:

perl bin/createsamples.pl positives.txt negatives.txt samples 1500\
  "opencv_createsamples -bgcolor 0 -bgthresh 0 -maxxangle 1.1\
  -maxyangle 1.1 maxzangle 0.5 -maxidev 40 -w 15 -h 15"

and trained like this:

opencv_traincascade -data classifier -vec samples.vec -bg negatives.txt\
  -numStages 20 -minHitRate 0.999 -maxFalseAlarmRate 0.5 -numPos 100\
  -numNeg 12 -w 15 -h 15 -mode ALL -precalcValBufSize 1024\
  -precalcIdxBufSize 1024

The classifier/cascade.xml is what i'm using with the facedetect.cpp sample from opencv: https://github.com/Itseez/opencv/blob/master/samples/c/facedetect.cpp

...I know I need a lot more sample images but I wanted to test something simple before I comited to all the cropping needed. Thanks for your help.

coin date cascade training

I have been trying to create a cascade classifier for a penny date prefix of the number 19. I am following this tutoria: http://coding-robin.de/2013/07/22/train-your-own-opencv-haar-classifier.html , but but I haven't had a positive result from detection. I am getting a circle in the middle regardless of what image i use.

C:\fakepath\opencv_penny_date_facedetect.png

Can someone please see if I am missing something?

Here are the files used for the training:

http://coinmatico.com/wp-content/uploads/2014/05/opencv-haar-classifier-training2.ziphttp://coinmatico.com/wp-content/uploads/2014/05/opencv-haar-classifier-training3.zip

created samples like this:

perl bin/createsamples.pl positives.txt negatives.txt samples 1500\
  "opencv_createsamples -bgcolor 0 -bgthresh 0 -maxxangle 1.1\
  -maxyangle 1.1 maxzangle 0.5 -maxidev 40 -w 15 -h 15"

and trained like this:

opencv_traincascade -data classifier -vec samples.vec -bg negatives.txt\
  -numStages 20 -minHitRate 0.999 -maxFalseAlarmRate 0.5 -numPos 100\
  -numNeg 12 -w 15 -h 15 -mode ALL -precalcValBufSize 1024\
  -precalcIdxBufSize 1024

The classifier/cascade.xml is what i'm using with the facedetect.cpp sample from opencv: https://github.com/Itseez/opencv/blob/master/samples/c/facedetect.cpp

...I know I need a lot more sample images but I wanted to test something simple before I comited to all the cropping needed. Thanks for your help.