Through this course research project, we aim to construct a fire detection system based on image processing techniques which can be implemented in existing surveillance devices like CCTV, wireless camera and UAVs.
We have chosen to implement our fire detection algorithm using the python programming
language. The Open Computer Vision library is utilized as it provides straightforward and
convenient methods for all the procedures in our algorithm. We generate a Custom HAAR
Cascade Classifier trained using 1600 images, obtained from ImageNet, in command line using OpenCV.
Meanwhile, we also create a python program to accept live video feed from an attached
camera. Built using Open Computer Vision functions, this script will be able to execute
operations on individual frames while still delivering real time results.
After obtaining the Custom Haar Cascade Classifier file, we integrate it into our python
program. The module compares each frame of the video using our classifier, and when a fire is
detected, a variety of alarms can be triggered, ranging from visual to audio.
