Fire incidents continue to pose serious risks to human safety, infrastructure, and the environment, especially in residential, industrial, and public settings where conventional systems primarily respond after a fire has already developed. This paper proposes an integrated Artificial Intelligence and Internet of Things based fire prediction and prevention system aimed at identifying early indicators of fire before ignition. The system combines environmental sensing with real-time visual analysis by utilizing sensors to monitor temperature, humidity, and combustible gases alongside camera-based observation of the surroundings. Both sensor readings and video data are analysed using intelligent algorithms to detect anomalies such as the presence of smoke, rapid temperature variations, and visual flame characteristics. A two-stage validation approach is employed, where sensor thresholds are evaluated in conjunction with visual detection outputs to estimate fire risk levels and reduce false positives. When a high-risk condition is identified, the system triggers local alert mechanisms and sends notifications to a remote monitoring interface to enable prompt intervention. By transitioning from conventional reactive detection methods to a predictive and preventive framework, the proposed system improves response efficiency and offers a scalable, cost-effective solution applicable to smart homes, commercial facilities, and industrial environments. Furthermore, the integration of multi-modal data sources enhances system robustness by enabling more accurate and reliable decision-making under diverse environmental conditions.
Mounika et al. (Mon,) studied this question.