Enhancing activity monitoring of smart homes using IOT sensors

Abstract

Accurate recognition of daily activities is vital for applications in healthcare, fitness tracking, and assisted living. Wrist-worn accelerometers, embedded in smartwatches and fitness trackers, provide an effective means of continuously monitoring user movements. Therefore this project explores the application of machine learning (ML) techniques to analyze accelerometer data from wrist-worn devices for precise daily activity recognition. The goal is to develop robust ML models that can process complex and noisy accelerometer data to identify and classify a wide range of activities such as walking, running, sitting, and sleeping. The proposed system utilizes advanced ML algorithms, to interpret the accelerometer data. These models are trained on extensive datasets containing labeled activity data, enabling them to learn intricate patterns and variations in user movements. The system's performance was rigorously tested and validated, demonstrating significant improvements in accuracy and reliability over traditional methods, such as manual logs, pedometers, and simple threshold-based algorithms. The ML-based approach offers comprehensive analysis and real-time monitoring capabilities, providing users with precise and continuous activity recognition. Personalized insights and recommendations are generated based on individual activity patterns, enhancing user experience and promoting better health outcomes. Additionally, the system supports assisted living by monitoring daily activities of the elderly and individuals with disabilities, ensuring their safety and well-being.