Abstract. The research is focused on enhancing the reliability of a fire detection system, which involves designing and demonstrating smart home fire detection and mobile alerting systems. This system increases fire detection sensitivity, reduces false alarms, and provides users with prompt alerts. The system incorporates more than one sensor, such as flame detection, an air quality sensor (measuring PM2.5, CO2, and TVOCs), temperature, and humidity levels, to enhance accuracy. Artificial intelligence algorithms filter real-time sensor streams and accurately identify real fire threats versus false alarms, thereby enabling more accurate alerts. The research was conducted in the municipality of President Roxas, North Cotabato, with outdoor testing for performance by thirty residents. Surveys, time trackers, and a system performance log data provider were then employed to measure detection accuracies, response times, and alert frequencies. Results showed a 7-15% increase in detection accuracy, a 50% decrease in response time, and a significant reduction in false alarms, resulting in an enhanced user experience. The statistical paired t-test comparisons also confirmed a significant difference in performance between the pre- and post-implementation periods. Findings suggest that AI-based fire detection systems can offer a viable solution for enhancing fire safety in homes and, consequently, reducing risks in the event of a fire.

Keywords: AI-powered fire detection; Detection accuracy; False alarm reduction; Home safety; Mobile notifications.