Dead End Data

Download dead_end.tar.gz and extract it into your home directory. Open a terminal and go into the directory.

$ cd ~/dead_end

Run the following command to generate the classifier. Note that training will take between 1-2 hours.

opencv_traincascade -data dead_end -vec samples.vec -bg negative_names.txt -numStages 20 -minHitRate 0.999 -maxFalseAlarmRate 0.5 -numPos 58 -numNeg 270 -w 48 -h 48 -mode ALL -precalcValBufSize 1024 -precalcIdxBufSize 1024

Once training is complete, move the classifier into the classifiers folder of this package. The classifier is located in the dead_end/dead_end directory.

$ cp ~/dead_end/dead_end/cascade.xml ~/<ros_ws>/src/sign_detection/classifiers/dead_end.xml

Edit the launch file to point to the new classifier

<param name="cascade_path" type="string" value="$(find sign_detection)/classifiers/dead_end.xml" />