SafeHer AI

Abstract

In particular, the lack of immediate human assistance is a major reason why women's safety continues to be a problem in society at large. Traditional safety apps mostly require the user to manually interact with them, so they are not helpful at all in the case of a sudden panic or physical distress. This article presents an AI-powered smart wristband system that can detect unsafe situations in real time and respond to emergencies automatically. The system under discussion uses wearable sensors such as an accelerometer-gyroscope module, a heart rate sensor, a microcontroller, a base embedded platform, and a smartphone interface. Wearable device motion data are used in this research work to locate abnormal or panic-like movement patterns by means of a Temporal Convolutional Network (TCN), whereas physiological stress can be detected on the basis of personalized heart rate analysis committed to individual baseline variations. Furthermore, through GPS-based risk zone analysis and time-dependent rules, the system's sensitivity could be dynamically changed, and hence achievement of contextual awareness. A multi-modal decision framework, which is based on user confirmation, is employed in order to minimize false alarm generation from bio signal interpretation. This mechanism is followed by the incorporation of physical features and personal health features coupled with motion that are then subjected to an analysis through a multi-modal decision framework, which results in the detection of the event and alarm generation. Once the system has figured out that the threat is real, it will not hesitate to send out an SOS alert as well as allow someone to track the live location of the person continuously through a cloud platform. This proposed model is a personalized, multi-sensor fusion and low-latency decision, approach that has been compared to a reliable and scalable solution for women.