Smartphone sensor-based human activity recognition using LSTM networks: Development and implementation in android applications
Loading...
Date
Publisher
Institute of Electrical and Electronics Engineers Inc.
Citation
D. Sikder, N. I. Rafin, M. G. R. Alam and M. Z. Uddin, "Smartphone Sensor-based Human Activity Recognition Using LSTM Networks: Development and Implementation in Android Applications," 2024 27th International Conference on Computer and Information Technology (ICCIT), Cox's Bazar, Bangladesh, 2024, pp. 203-208, doi: 10.1109/ICCIT64611.2024.11021865.
Abstract
The machine learning approach to estimate human activity using smartphone sensor data is challenging. In this work, a Human Activity Recognition (HAR) approach is conducted based on the Long Short-Term Memory (LSTM) model, which can recognize six different behaviors: Downstairs, Jogging, Sitting, Standing, Upstairs, and Walking. To achieve the best potential result, various machine learning and statistical approaches were explored. The LSTM model, known for its effectiveness in sequence prediction, was chosen for its ability to run efficiently on lightweight edge devices such as smartphones. This model achieved a test accuracy of 97%. Finally, the model was exported and deployed in an Android application, providing a user-friendly interface.
Keywords
Legged locomotion, Deep learning, Recurrent neural networks, Accuracy, Computational modeling, Predictive models, Behavioral sciences, Human activity recognition, Information technology, Long short term memory, Machine Learning (ML), Deep learning, Human activity (HA), Human activity recognition (HAR), TensorFlow, Recurrent neural networks (RNNs), Long short-term memory networks (LSTM)
LC Subject Headings
Description
Publisher Link
Type
Conference Proceeding