Modeling complex human activities with inertial units, ambient sensors, and machine learning

• Laura Sthefany Colombo UNESP
• André Luiz da Silva Conde UNESP
• Leonardo Tórtoro Pereira UNESP
• Caetano Mazzoni Ranieri UNESP

DOI: https://doi.org/10.5753/eniac.2025.14517

Abstract

Human activity recognition is key in applications such as healthcare, sports, and smart environments. Recent studies apply both classical algorithms and deep learning models. However, few address the impact of similar and complex activity patterns on model performance. This work compares Decision Trees, Random Forests, and a 1D Convolutional Neural Network (CNN-1D) using two public datasets: PAMAP2 and HWU-USP. On PAMAP2, the CNN1D proved most robust (62.48% ± 10.31% accuracy), while on HWU-USP the Random Forest was more stable (38.93% ± 3.94% accuracy), highlighting the importance of hyperparameter tuning for complex tasks.

References

Alagoz, C. (2024). Comparative analysis of xgboost and minirocket algortihms for human activity recognition. arXiv preprint arXiv:2402.18296.

Cicirelli, G., Marani, R., Petitti, A., Milella, A., and D’Orazio, T. (2021). Ambient assisted living: A review of technologies, methodologies and future perspectives for healthy aging of population. Sensors, 21(10).

Dempster, A., Schmidt, D. F., and Webb, G. I. (2021). Minirocket: A very fast (almost) deterministic transform for time series classification. In Proceedings of the 27th ACM SIGKDD conference on knowledge discovery & data mining, pages 248–257.

D’Arco, L., Wang, H., and Zheng, H. (2022). Assessing impact of sensors and feature selection in smart-insole-based human activity recognition. Methods and Protocols, 5(3):45.

Guo, W., Yamagishi, S., and Jing, L. (2024). Human activity recognition via wi-fi and inertial sensors with machine learning. IEEE Access, 12:18821–18836.

Jovanovic, M., Mitrov, G., Zdravevski, E., Lameski, P., Colantonio, S., Kampel, M., Tellioglu, H., and Florez-Revuelta, F. (2022). Ambient assisted living: Scoping review of artificial intelligence models, domains, technology, and concerns. J Med Internet Res, 24(11):e36553.

Manouchehri, N. and Bouguila, N. (2023). Human activity recognition with an hmm-based generative model. Sensors, 23(3):1390.

Marques, G., Bhoi, A. K., Albuquerque, V., and S., H. (2021). IoT in Healthcare and Ambient Assisted Living. Springer.

O’Malley, T., Bursztein, E., Long, J., Chollet, F., Jin, H., Invernizzi, L., et al. (2019). Keras Tuner. [link].

Ranieri, C. M., MacLeod, S., Dragone, M., Vargas, P. A., and Romero, R. A. F. (2021). Activity recognition for ambient assisted living with videos, inertial units and ambient sensors. Sensors, 21(3):768.

Reiss, A. and Stricker, D. (2012). Introducing a new benchmarked dataset for activity monitoring. In 2012 16th international symposium on wearable computers, pages 108–109. IEEE.

Yadav, S. K., Tiwari, K., Pandey, H. M., and Akbar, S. A. (2021). A review of multimodal human activity recognition with special emphasis on classification, applications, challenges and future directions. Knowledge-Based Systems, 223:106970.