Fire and Smoke Detection System

Fire and smoke detection is a critical component of early warning systems aimed at preventing the rapid spread of fires in both urban and forest environments. Traditional detection methods, such as thermal sensors and manual monitoring, often suffer from limitations including delayed response times, environmental dependencies, and high maintenance costs. In this study, we propose a real-time fire and smoke detection system utilizing the YOLOv8 deep learning model. The system is trained on a dataset of over 11,000 annotated images sourced from Roboflow, enabling high-accuracy detection in diverse scenarios. The proposed method incorporates a structured pipeline comprising image preprocessing, normalization, model training, and alert generation. Experimental results demonstrate that the YOLOv8 model achieves a mean average precision (mAP@50) of 0.926, with a precision of 0.954 and recall of 0.848, outperforming previous YOLO variants. This system offers a robust solution for real-time fire and smoke detection, with applications in forest surveillance, urban safety, and industrial monitoring.