System For Real Time Fire And Smoke Intensity Detection.

Fire poses a significant threat to daily life, causing both economic and social harm. To mitigate these damages, early detection of fire and smoke is crucial, and this paper introduces a model employing vision-based techniques. The proposed model utilizes image processing and convolutional neural networks to detect fire and smoke, providing insights into their intensity and any changes in a video. The model comprises two units for fire and smoke detection, each employing image preprocessing techniques, including rule-based color detection and motion detection, along with CNN. The calculated percentages of fire and smoke in the processed images offer detailed information about the severity of the hazards in a specific area. The model detects whether the intensity of fire and smoke is increasing, decreasing or constant. [ABSTRACT FROM AUTHOR]