Real-Time Acoustic Scene Recognition for Elderly Daily Routines Using Edge-Based Deep Learning

Sensors, Journal Year: 2025, Volume and Issue: 25(6), P. 1746 - 1746

Published: March 12, 2025

The demand for intelligent monitoring systems tailored to elderly living environments is rapidly increasing worldwide with population aging. Traditional acoustic scene that rely on cloud computing are limited by data transmission delays and privacy concerns. Hence, this study proposes an recognition system integrates edge deep learning enable real-time of individuals’ daily activities. consists low-power devices equipped multiple microphones, portable wearable components, compact power modules, ensuring its seamless integration into the lives elderly. We developed four models—convolutional neural network, long short-term memory, bidirectional network—and used model quantization techniques reduce computational complexity memory usage, thereby optimizing them meet device constraints. CNN demonstrated superior performance compared other models, achieving 98.5% accuracy, inference time 2.4 ms, low requirements (25.63 KB allocated Flash 5.15 RAM). This architecture provides efficient, reliable, user-friendly solution in care.

Language: Английский

Real-Time Prediction of Residents’ ADL using Asynchronous Multivariables Time-series Signals

2021 IEEE International Conference on Big Data (Big Data), Journal Year: 2024, Volume and Issue: unknown, P. 8697 - 8699

Published: Dec. 15, 2024

Language: Английский