Fire And Smoke Detection

Fire stands out as one of the most dangerous disasters to human life, property, and ecology, and it usually develops very fast due to late detection. Classic fire detection technologies, such as heat sensors and smoke alarms, provide very limited information and mostly fail to respond in the early stage of a fire. In this paper, the authors propose a real-time Deep Learning Based Detection of Fire and smoke-based YOLO model to handle the aforementioned issues. It takes continuous video streams from a webcam or CCTV and detects fire and smoke in them with high accuracy. Each frame is pre-processed before feeding into the model, which outlines the regions, classification tags, and confidence metrics. Instant alerts will be sent once fire or smoke has been detected to assure timely intervention. The efficacy of the proposed approach in real-time on regular hardware makes it much more reliable compared to normal sensor-based methods. Experimental results prove that the proposed system is robust, accurate, and suitable for practical applications in safety monitoring.