YOLO-BASED ABNORMAL BEHAVIOR DETECTION SYSTEM FOR ELDERLY HEALTHCARE MONITORING

Abstract

The YOLO-Based Abnormal Behavior Detection System for Elderly Healthcare Monitoring presents an intelligent and predictive approach to enhancing the safety and well-being of elderly individuals. The system integrates computer vision, machine learning, and embedded sensing technologies to monitor both physical activity and physiological parameters in real time. Using a Python-based embedded processor, the YOLO (You Only Look Once) algorithm analyzes live video captured via a USB web camera to detect abnormal behaviors such as falls, sudden collapses, or prolonged inactivity. Machine learning models further process the collected behavioral and physiological data to predict potential health risks and emergencies before they occur. Simultaneously, an Arduino microcontroller interfaces with heartbeat and temperature sensors to continuously track vital signs. Upon detection of unusual activity, critical health conditions, or predicted emergencies, the system triggers a buzzer for immediate local alerts and transmits notifications to caregivers or family members through a GSM module. Powered by a regulated 12V supply and interconnected via secure connectors, this system offers a reliable, real-time, and predictive monitoring solution. By combining behavioral analysis, vital sign monitoring, and machine learning-based predictions, the system enables proactive interventions, reducing emergency response time and improving elderly care management