안전 관리 애플리케이션 구현 = Implementation of Safety Management Application

다국어 초록 (Multilingual Abstract)

In this paper, we propose a smart safety management application that integrates advanced AI technology with smartphone sensors to effectively respond to fall accidents frequently occurring at industrial sites. The primary objective of this study is to protect workers' safety in real-time and minimize casualties by establishing a rapid rescue system in the event of an accident. The proposed system provides two distinct fall detection mechanisms depending on the user's environment. First, when a worker carries a smartphone, it analyzes physical movements using the built-in accelerometer and gyroscope sensors. Second, when a smartphone is fixed at a site, it supports high-precision fall detection without specialized equipment by accurately inferring the worker's posture through the MoveNet model. Upon detecting a fall accident, the system activates a multi-stage notification process— connecting managers, fellow workers, and external emergency relief agencies—to prevent secondary accidents and secure the "golden hour." Operating in an on-device environment, the proposed system is cost-effective and easy to implement even in small-scale workplaces. It is expected to contribute to the construction of a social safety net by expanding into fields such as elderly care and medical monitoring in the future.