Enhancing Human Activity Recognition (HAR) via Semi-Adaptive Sliding Windows with Triplet Subwindow Similarity and Growth Capping

In recent years, sensor-based human activity recognition has become an active research topic. Sensor data are typically represented as time series, which require a segmentation process before feature extraction and machine learning–based classification. The sliding window is one of the most commonly used segmentation techniques; however, determining the optimal window length remains a major challenge for achieving accurate activity recognition performance. This study proposes a semi-adaptive sliding window method that integrates static and dynamic strategies using accelerometer sensor data. The proposed approach exploits temporal information by considering the current, past, and future windows, each of which is further divided into three sub-windows. The window size is adaptively updated based on the similarity among pairs of subwindows forming a triplet subwindow, and a growth-capping mechanism is incorporated to prevent excessive window expansion. Performance evaluation was conducted using the XGBoost and LightGBM classifiers on the FORTH-TRACE, SBHARPT, WISDM, and PAMA2 datasets. The experimental results show that using XGBoost and LightGBM, the proposed method achieves accuracies of 97.26% and 97.26% on the FORTH-TRACE dataset, 98.09% and 98.15% on the SBHARPT dataset, 98.97% and 99.06% on the WISDM dataset, and 92.21% and 92.43% on the PAMA2 dataset, respectively. These results demonstrate that the proposed semi-adaptive sliding window approach consistently improves human activity recognition performance.

Accelerometer; Activity-recognition; Semi-adaptive; Similarity; Subwindow

Akter, M., Ansary, S., Khan, M. A. M., & Kim, D. (2023). Human activity recognition using attention-mechanism-based deep learning feature combination. Sensors, 23(12), 5715. https://doi.org/10.3390/S23125715

Baraka, A. M. A., & Mohd Noor, M. H. (2023). Similarity segmentation approach for sensor-based activity recognition. IEEE Sensors Journal, 23(17), 19704–19716. https://doi.org/10.1109/JSEN.2023.3295778

Baraka, A. R., & Mohd Noor, M. H. (2024). Deep similarity segmentation model for sensor-based activity recognition. Multimedia Tools and Applications, 1–24. https://doi.org/10.1007/S11042-024-18933-2/METRICS

Chandramouli, N. A., Natarajan, S., Alharbi, A. H., Kannan, S., Khafaga, D. S., Raju, S. K., Eid, M. M., & El-kenawy, E. S. M. (2024). Enhanced human activity recognition in medical emergencies using a hybrid deep CNN and bi-directional LSTM model with wearable sensors. Scientific Reports, 14(1), 1–24. https://doi.org/10.1038/s41598-024-82045-y

Chen, S. Y., & Lin, C. L. (2024). Wi-fi-based human activity recognition for continuous, whole-room monitoring of motor functions in Parkinson’s disease. IEEE Open Journal of Antennas and Propagation, 5(3), 788–799. https://doi.org/10.1109/OJAP.2024.3393117

Dilshad Ansari, M., Benhaili, Z., Visutsak, P., Bangkok, N., Ashaq Hussain Bhat, T., Noori, F. M., Deeptha, R., Ramkumar, K., Venkateswaran, S., Mehedi Hassan, M., Rafiul Hassan, M., & Zia Uddin, M. (2024). Enhancing human activity recognition for the elderly and individuals with disabilities through optimized internet-of-things and artificial intelligence integration with advanced neural networks. Frontiers in Neuroinformatics, 18.

Guerra, B. M. V., Torti, E., Marenzi, E., Schmid, M., Ramat, S., Leporati, F., & Danese, G. (2023). Ambient assisted living for frail people through human activity recognition: State-of-the-art, challenges and future directions. Frontiers in Neuroscience, 17, 1256682. https://doi.org/10.3389/FNINS.2023.1256682

Hirawat, A., Taterh, S., & Sharma, T. K. (2022). A dynamic window-size based segmentation technique to detect driver entry and exit from a car. Journal of King Saud University - Computer and Information Sciences, 34(10), 8514–8522. https://doi.org/10.1016/J.JKSUCI.2021.08.028

Host, K., & Ivaši?-Kos, M. (2022). An overview of human action recognition in sports based on computer vision. Heliyon, 8(6), e09633. https://doi.org/10.1016/J.HELIYON.2022.E09633

Jaén-Vargas, M., Leiva, K. M. R., Fernandes, F., Goncalves, S. B., Silva, M. T., Lopes, D. S., & Olmedo, J. J. S. (2022). Effects of sliding window variation in the performance of acceleration-based human activity recognition using deep learning models. PeerJ Computer Science, 8, e1052. https://doi.org/10.7717/PEERJ-CS.1052

Jeon, S., Lee, Y. S., & Son, S. H. (2023). Cascade windows-based multi-stream convolutional neural networks framework for early detecting in-sleep stroke using wristbands. IEEE Access, 11, 84944–84956. https://doi.org/10.1109/ACCESS.2023.3301872

Karagiannaki, K., Panousopoulou, A., & Tsakalides, P. (2016). A benchmark study on feature selection for human activity recognition. Proceedings of the 2016 ACM International Joint Conference on Pervasive and Ubiquitous Computing, 105–108. https://doi.org/10.1145/2968219.2971421

Liandana, M., Hostiadi, D. P., Hendrawan, N. R., Pradipta, G. A., Desiana, P., & Ayu, W. (2025). Enhanced human activity recognition (HAR): Leveraging sub-window techniques and feature ratios from triaxial accelerometer data. International Journal of Intelligent Engineering and Systems, 18(1). https://doi.org/10.22266/ijies2025.0229.34

Liandana, M., Hostiadi, D. P., & Pradipta, G. A. (2024). A new approach for human activity recognition (HAR) using a single tri-axial accelerometer based on a combination of three feature subsets. International Journal of Intelligent Engineering and Systems, 17(2), 235–250. https://doi.org/10.22266/ijies2024.0430.21

Lu, N., Yan, T., Zhu, S., Qian, J., & Han, M. (2025). Deep feature unsupervised domain adaptation for time-series classification. IEEE Transactions on Artificial Intelligence, 6(3), 725–737. https://doi.org/10.1109/TAI.2024.3491948

Ma, C., Li, W., Cao, J., Du, J., Li, Q., & Gravina, R. (2020). Adaptive sliding window based activity recognition for assisted livings. Information Fusion, 53, 55–65. https://doi.org/10.1016/J.INFFUS.2019.06.013

Machado, J., Antosz, K., Mazurkiewicz, D., Ren, Y., Rea, P., El Abdi, R., Ranga, M., Kumar Manupati, V., Villani, E., & Park, K. (2022). Wearable sensor for forearm motion detection using a carbon-based conductive layer-polymer composite film. Sensors, 22(6), 2236. https://doi.org/10.3390/S22062236

Mekruksavanich, S., & Jitpattanakul, A. (2022). Deep residual network for smartwatch-based user identification through complex hand movements. Sensors, 22(8), 3094. https://doi.org/10.3390/S22083094

Minh Dang, L., Min, K., Wang, H., Jalil Piran, M., Hee Lee, C., & Moon, H. (2020). Sensor-based and vision-based human activity recognition: A comprehensive survey. Pattern Recognition, 108, 107561. https://doi.org/10.1016/J.PATCOG.2020.107561

Munoz-Organero, M., Luptáková, I. D., Kubov?ík, M., & Pospíchal, J. (2022). Wearable sensor-based human activity recognition with transformer model. Sensors, 22(5), 1911. https://doi.org/10.3390/S22051911

Pan, J., Hu, Z., Zhang, L., & Cai, X. (2025). Multi-channel time series decomposition network for generalizable sensor-based activity recognition. IEEE Transactions on Automation Science and Engineering, 22, 8150–8161. https://doi.org/10.1109/TASE.2024.3480119

Pathan, M. S., Nag, A., Pathan, M. M., & Dev, S. (2022). Analyzing the impact of feature selection on the accuracy of heart disease prediction. Healthcare Analytics, 2, 100060. https://doi.org/10.1016/J.HEALTH.2022.100060

Petz, P., Eibensteiner, F., & Langer, J. (2021). Sensor shirt as universal platform for real-time monitoring of posture and movements for occupational health and ergonomics. Procedia Computer Science, 180, 200–207. https://doi.org/10.1016/J.PROCS.2021.01.157

Qian, H., Pan, S. J., & Miao, C. (2021). Weakly-supervised sensor-based activity segmentation and recognition via learning from distributions. Artificial Intelligence, 292, 103429. https://doi.org/10.1016/J.ARTINT.2020.103429

Rahayu, Y., Rosdiansyah, Hilmi, M. F., & Odih, T. (2021). Wearable antenna for time-domain breast tumor detection. International Journal of Technology, 12(6), 1101–1111. https://doi.org/10.14716/IJTECH.V12I6.5187

Reiss, A., & Stricker, D. (2012). Introducing a new benchmarked dataset for activity monitoring. Proceedings - International Symposium on Wearable Computers, ISWC, 108–109. https://doi.org/10.1109/ISWC.2012.13

Reyes-Ortiz, J. L., Oneto, L., Samà, A., Parra, X., & Anguita, D. (2016). Transition-aware human activity recognition using smartphones. Neurocomputing, 171, 754–767. https://doi.org/10.1016/J.NEUCOM.2015.07.085

Rustam, F., Reshi, A. A., Ashraf, I., Mehmood, A., Ullah, S., Khan, D. M., & Choi, G. S. (2020). Sensor-based human activity recognition using deep stacked multilayered perceptron model. IEEE Access, 8, 218898–218910. https://doi.org/10.1109/ACCESS.2020.3041822

Shi, W., Fang, X., Yang, G., & Huang, J. (2022). Human activity recognition based on multi-channel convolutional neural network with data augmentation. IEEE Access, 10, 76596–76606. https://doi.org/10.1109/ACCESS.2022.3192452

Sun, L., Yang, X., & Hu, C. (2022). DSW-HAR: A dynamic sliding window based human activity recognition method. Proceedings - 2022 IEEE SmartWorld, Ubiquitous Intelligence and Computing, Autonomous and Trusted Vehicles, Scalable Computing and Communications, Digital Twin, Privacy Computing, Metaverse, 1421–1426. https://doi.org/10.1109/SMARTWORLD-UIC-ATC-SCALCOM-DIGITALTWIN-PRICOMP-METAVERSE56740.2022.00205

Vidya, B., & Sasikumar, P. (2022). Wearable multi-sensor data fusion approach for human activity recognition using machine learning algorithms. Sensors and Actuators A: Physical, 341, 113557. https://doi.org/10.1016/J.SNA.2022.113557

Wang, H., Zhao, J., Li, J., Tian, L., Tu, P., Cao, T., An, Y., Wang, K., & Li, S. (2020). Wearable sensor-based human activity recognition using hybrid deep learning techniques. Security and Communication Networks, 2020. https://doi.org/10.1155/2020/2132138

Wang, J., Xu, P., Ji, X., Li, M., & Lu, W. (2023a). Feature selection in machine learning for perovskite materials design and discovery. Materials, 16(8), 3134. https://doi.org/10.3390/MA16083134

Wang, K., Saragadam, A., Kaur, J., Dogra, A., Cao, S., Ghafurian, M., Butt, Z. A., Abhari, S., Chumachenko, D., & Morita, P. P. (2025). A contactless method for recognition of daily living activities for older adults based on ambient assisted living technology. Internet of Things, 30, 101502. https://doi.org/10.1016/J.IOT.2025.101502

Wang, L., Jiang, S., & Jiang, S. (2021). A feature selection method via analysis of relevance, redundancy, and interaction. Expert Systems with Applications, 183, 115365. https://doi.org/10.1016/J.ESWA.2021.115365

Wang, Y., Xu, H., Liu, Y., Wang, M., Wang, Y., Yang, Y., Zhou, S., Zeng, J., Xu, J., Li, S., & Li, J. (2023b). A novel deep multifeature extraction framework based on attention mechanism using wearable sensor data for human activity recognition. IEEE Sensors Journal, 23(7), 7188–7198. https://doi.org/10.1109/JSEN.2023.3242603

Wang, Y. H., Zhang, Y. F., Zhang, Y., Gu, Z. F., Zhang, Z. Y., Lin, H., & Deng, K. J. (2022). Identification of adaptor proteins using the ANOVA feature selection technique. Methods, 208, 42–47. https://doi.org/10.1016/J.YMETH.2022.10.008

WISDM Lab. (n.d.). Dataset [Retrieved September 21, 2024, from https://www.cis.fordham.edu/wisdm/dataset.php].

Yu, X., & Al-Qaness, M. A. A. (2025). Human activity recognition using deep residual convolutional network based on wearable sensors. IEEE Journal of Biomedical and Health Informatics, 29(3), 1950–1958. https://doi.org/10.1109/JBHI.2024.3510860

Zhang, C., Cao, K., Lu, L., & Deng, T. (2022). A multi-scale feature extraction fusion model for human activity recognition. Scientific Reports, 12(1), 1–13. https://doi.org/10.1038/s41598-022-24887-y