AI–POWEREDSMARTFIREDETECTIONFROMCCTV USING DEEP LEARNING

Abstract

When fire breaks out in places like building, shopping malls, colleges, schools and industries, most of the time, this leads to serious damages, as they are not spotted at an early stage. The standard systems of fire detection, such as smoke and heat sensors tend to be responsive after a long time and may also give false alarms. To alleviate these shortcomings, this paper will propose a smart fire detection systems which involves the use of computer vision and deep learning algorithms to detect fire at an early stage using camera footage of a CCTV camera. The proposed system will be based on the YOLOv8 deep learning and OpenCV to analyze the video streams in real-time and identify any visuals related to fire (flame and smoke). This system does not require physical sensors and instead uses the already existing surveillance cameras which makes the system very cost efficient and simple to install. After a fire is identified with high confidence, an automatic notification is forwarded to the emergency department via email, which will aid in prompt response and minimise the use of human surveillance. The system is created on the basis of such technologies as python, Django framework and MySQL database to control the video processing, the results of video processing, and alerts. According to the experimental results, the system is effective and has an accuracy of about 92 percent with a rapid response time in various indoor environmental conditions. On the whole, this piece of writing shows that the regular CCTV systems CCTV systems may be converted to intelligent fire monitoring systems, helping to enhance safety and minimize the risk of fires.

Keywords