cigarette-detection-with-haar-cascade-classifier's Introduction

Cigarette Detection

Building on Linux
Dataset
Preparation
Training
Results

Building on Linux

We are going to follow the instructions given here:

To build you need CMake and OpenCV on your system.

$ git clone https://github.com/Itseez/opencv.git

Compiler:

sudo apt-get install build-essential

Required Libraries:

$ sudo apt-get install cmake git libgtk2.0-dev pkg-config libavcodec-dev libavformat-dev libswscale-dev

Python bindings and such:

$ sudo apt-get install python-dev python-numpy libtbb2 libtbb-dev libjpeg-dev libpng-dev libtiff-dev libjasper-dev libdc1394-22-dev

OpenCV development library:

sudo apt-get install libopencv-dev

Dataset

As positive image, I tried both training the haar cascade with a cigarette image (cigar100x75.jpg) and a hand image holding a cigarette (cigar100x84.jpg). As negative image, I took a sample of 5394 images from a face dataset.

Preparation

We need to create files which contains information about our dataset.

Run create_neg() and create_pos() functions in the sample_creator.ipynb notebook to create those files named as "bg.txt" which contains the paths of negative images and "info.dat" which contains the coordinates of the cigarette object in our positive image.

After that, run the following commands in terminal to create positive samples:

opencv_createsamples -img cigar100x84.jpg -bg bg.txt -info positive_images/info.lst -pngout positive_images -maxxangle 0.5 -maxyangle 0.5 -maxzangle 0.5 -num 1950

Once the positive samples have been created, we need a new file which contains information stating that where the cigarette is in our recently created positive images, just like "info.dat". We will call it as "info.lst"

opencv_createsamples -info positive_images/info.lst -num 1950 -w 50 -h 40 -vec positives.vec

Training

Now, it's time to train a haar cascade to detect cigarettes in images and hopefully in videos.

opencv_traincascade -data data -vec positives.vec -bg bg.txt -numPos 1900 -numNeg 950 -numStages 6 -w 50 -h 40

Done!

Results

Now we can try our cascade using the codes in the detect_my_cigar.ipynb notebook.

cigarette-detection-with-haar-cascade-classifier's People

paradiddle131 / cigarette-detection-with-haar-cascade-classifier Goto Github PK