Many rooftop fire suppression systems malfunction or spray too much water in the wrong place. Leading to a lot of damage being done which can be very expensive to fix.

Our system goes through multiple steps to identify and spray water at the fire.

• The laptop captures a live video feed. The JavaScript code scans every single pixel in the frame per second. We also programmed a ratio filter, instead of just looking for red,the AI checks if the value red (R) is at least 2.2 times stronger than the values green (G) and blue (B).
• Becuase fire is a light source, the code checks if the center of the fire is very bright (almost white or very light yellow) to be able to tell if the fire is real or fake (like a drawing or something).
• Another thing for our system to be able to tell if the fire is real or fake is a temporal analysis function. What this does is the code compares the current image with the previous one. If the number of red pixels change rapidly, the AI classifies the threat as a real fire. If the number of red pixels is stable then it is classified as a fake fire.
• Once the computer has decided where the fire is, it sends a message via the USB cable.
• When the Arduino receives the coordinates, it uses the function ‘map()’ to transform them into degrees that the servos can understand. Once the servos get the degrees the turn towards the fire. The Arduino activates the digital pins (4 or 7 ) to turn on the corresponding light (blue=safe, red=fire detected).

(this is an innovation project) the performance of our system had to be analysed and tweaked to make it function properly and meet our goals.This is basically it cuz our project is innovative.

We have developed a prototype monitoring system that uses code and a camera to detect and locate fires. Unlike conventional detectors, our system uses computer vision to distinguish a fire from a stationary red object, thus reducing false alarms while automatically directing the servos to the source of the danger to spray at the fire.

• https://docs.arduino.cc/hardware/uno-rev3/
• https://www.target-fire.co.uk/
• https://projecthub.arduino.cc/
• JavaScript.info
• https://forum.arduino.cc/

we would like to acknowledge our dad Amr and our mom Hala for support and always believing in us in difficult times of our project and always being there to support us.