README.md

Machine learning Steps

1. List positive and negative image in text file

   • Linux

   find ./positive_images -iname "*.jpg" > positives.txt
   find ./negative_images -iname "*.jpg" > negatives.txt

   • Windows

   forfiles /s /m *.jpg /c "cmd /c echo @relpath" > positives.txt
   forfiles /s /m *.jpg /c "cmd /c echo @relpath" > negatives.txt

   ** Remove quotes in the files.**

    ** Replace .\ -> ./positive_images/ **  
   

2. Create samples

perl ML/tools/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 50 -h 70"

3. Create .vec files from samples

python ML/tools/mergevec.py -v samples/ -o samples.vec

4. Train

opencv_traincascade -data classifier -vec samples.vec -bg negatives.txt -numStages 20 -minHitRate 0.999 -maxFalseAlarmRate 0.5 -numPos 1000 -numNeg 5 -w 50 -h 70 -mode ALL -precalcValBufSize 1024 -precalcIdxBufSize 1024

More information here :

• opencv-haar-classifier-training
• Train your own OpenCV Haar Classifier