A Deep Convolutional LSTM for ADLs Classification of the Elderly
Abstract:

Activity analysis systems or activity recognition systems for the elderly is recently a part of the smart home systems design. This assisted system normally helps the elderly to live alone in a house, safely and improve a quality of life. Therefore, learning to recognize which activities are safe is necessary for classifying the activity of the elderly. Furthermore, this information will give us some insights to understand the basic daily lives of the elderly and it also helps us to monitor activity and health information of the elderly. In this paper, we collected activities data using the multi-sensor motion sensors embedded inside the smartwatch (Fitbit). We also present the novel method for detecting and recognizing the activity using Deep Convolutional LSTM. In brief, this paper shows that the proposed method yields 88.425% of accuracy for activity classification. The paper also compares the results with our previous work which used Backpropagation Neural Networks as a classifier (78% of accuracy).
A Deep Convolutional LSTM for ADLs Classification of the Elderly
Ref.
P. Vanijkachorn and P. Visutsak, "A Deep Convolutional LSTM for ADLs Classification of the Elderly," 2021 International Conference on Data Analytics for Business and Industry (ICDABI), Sakheer, Bahrain, 2021, pp. 124-128, doi: 10.1109/ICDABI53623.2021.9655856.
keywords: {Backpropagation;Training;Smoothing methods;Convolution;Neural networks;Low-pass filters;Smart homes;Elderly;ADLs;Opportunity;ConvLSTM;Wearable device;Health condition;COVID-19},
Comments