Synthetic Sensor Data Generation for Activity Recognition

Ishwarya K. 1, N. Sowmya 2, A. Alice Nithya3, Nandhini G. 4 and D. Devi 5
1 Department of Computing Technologies, School of Computing, SRM Institute of Science and Technology Kattankulathur – 603 203
2 Department of Electronics and Communication Engineering. , SRM Institute of Science and Technology, Kattankulathur, 603203, Tamilnadu,India
3 Department of Computational Intelligence, School of Computing,SRM Institute of Science and Technology, Kattankulathur, 603203, Tamilnadu,India
3 Department of Computational Intelligence, School of Computing,SRM Institute of Science and College,West Tambaram poonthandalam,Chennai -602109
5Department of Computer Science and Engineering , Sathyabama institute of science and technology,Jeppiaar Nagar, Chennai-600 119.
ishwaryk3@srmist.edu.in
sowmyan1@srmist.edu.in*
alicenia@srmist.edu.in
devi.cse@sathyabama.ac.in
nandhini.am@sairam.edu.in
DOI: 10.46793/BISEC25.326I

 

ABSTRACT: Human Activity Recognition (HAR) is essential for a variety of appli- cations, including assisted living, fitness tracking, and healthcare monitoring. The accelerometer, gyroscope, and magnetometer—three inertial sensors that are built into smartphones—are used in this paper’s hybrid deep learning framework for sensor-based HAR. The raw sensor data goes through a thorough preparation step that includes mean/mode replacement techniques for handling missing val- ues and Kalman Filter identification for outliers. For efficient feature extraction, the cleaned data is subsequently fed into a Modified One-Dimensional Convolutional Neural Network (1D-CNN), which captures local temporal relationships in sensor signals. A Long Short-Term Memory (LSTM) network is then fed this data in order to identify intricate human actions and learn long-range temporal patterns. Real-time, precise activity prediction is possible with the suggested method, opening the door for context-aware applications in ubiquitous computing settings. Examining the possibility of using transfer learning strategies to modify the created models for use in new activity recognition tasks or data-poor domains. Models are being optimised for low-power and resource-constrained environments, particularly for wearable technology, to guarantee prolonged use without using excessive amounts of energy.

KEYWORDS: Kalman Filter, Convolutional Neural Network, Long Short-Term Memory , Temporal patterns.

REFERENCES:

  1. Wu, Jingwen, et al. “Detecting the quantitative hydrological response to changes in climate and human activities.” Science of the Total Environment 586 (2017): 328-337.
  2. Xu, Yang, et al. “Propagation from meteorological drought to hydrological drought under the impact of human activities: A case study in northern China.” Journal of Hydrology 579 (2019): 124147.
  3. Ni, Bingbing, et al. “Multilevel depth and image fusion for human activity detec- tion.” IEEE transactions on cybernetics 43.5 (2013): 1383-1394.
  4. Ke, Shian-Ru, et al. “A review on video-based human activity recognition.” Computers 2.2 (2013): 88-131.
  5. Jiménez‐Bonilla, Alejandro, Manuel Díaz‐Azpiroz, and Miguel Rodríguez Rodríguez. “Tectonics may affect closed watersheds used to monitor climate change and human activity effects.” Terra Nova 35.1 (2023): 58-65.
  6. Ronao, Charissa Ann, and Sung-Bae Cho. “Human activity recognition with smartphone sensors using deep learning neural networks.” Expert systems with applications 59 (2016): 235-244.
  7. Csurka, Gabriela. “Domain adaptation for visual applications: A comprehensive survey.” arXiv preprint arXiv:1702.05374 (2017).
  8. Jain, Yashswi, et al. “PoseCVAE: Anomalous human activity detection.” 2020 25th Inter- national Conference on Pattern Recognition (ICPR). IEEE, 2021.
  9. Zhang, Ying, et al. “Vegetation dynamics and its driving forces from climate change and human activities in the Three-River Source Region, China from 1982 to 2012.” Science of the Total Environment 563 (2016): 210-220.
  10. Piergiovanni, A. J., et al. “Adversarial generative grammars for human activity predic- tion.” Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part II 16. Springer International Publishing, 2020.
  11. Mukherjee, Debadyuti, et al. “EnsemConvNet: a deep learning approach for human activ- ity recognition using smartphone sensors for healthcare applications.” Multimedia Tools and Applications 79 (2020): 31663-31690.
  12. Javed, Abdul Rehman, et al. “A smartphone sensors-based personalized human activity recognition system for sustainable smart cities.” Sustainable Cities and Society 71 (2021): 102970.
  13. Shoaib, Muhammad, et al. “Complex human activity recognition using smartphone and wrist-worn motion sensors.” Sensors 16.4 (2016): 426.
  14. Li, Haobo, et al. “Bi-LSTM network for multimodal continuous human activity recognition and fall detection.” IEEE Sensors Journal 20.3 (2019): 1191-1201.
  15. Hassan, Mahmoud O., and Yasser M. Hassan. “Effect of human activities on floristic com- position and diversity of desert and urban vegetation in a new urbanized desert ecosys- tem.” Heliyon 5.8 (2019).
  16. Ignatov, Andrey, et al. “Ai benchmark: Running deep neural networks on android smartphones.” Proceedings of the European Conference on Computer Vision (ECCV) Workshops. 2018.
  17. Almaslukh, Bandar, Jalal Al Muhtadi, and Abdel Monim Artoli. “A robust convolutional neural network for online smartphone-based human activity recognition.” Journal of Intelli- gent & Fuzzy Systems 35.2 (2018): 1609-1620.
  18. Tsinganos, Kanaris, et al. “ERA-PLANET, a European Network for Observing Our Chang- ing Planet.” Sustainability 9.6 (2017): 1040.
  19. Kautz, Thomas, et al. “Activity recognition in beach volleyball using a Deep Convolutional Neural Network: Leveraging the potential of Deep Learning in sports.” Data Mining and Knowledge Discovery 31 (2017): 1678-1705.
  20. Saunois, Marielle, et al. “The growing role of methane in anthropogenic climate change.” Environmental Research Letters 11.12 (2016): 120207.
  21. Ishwarya, K., and A. Alice Nithya. “Performance-enhanced real-time lifestyle tracking model based on human activity recognition (PERT-HAR) model through smartphones.” The Journal of Supercomputing 78.4 (2022): 5241-5268.
  22. Nithya, A. Alice, et al. “CNN based Identifying Human Activity using Smartphone Sen- sors.” 2022 International Conference on Edge Computing and Applications (ICECAA). IEEE, 2022.
  23. Kumar, Pranjal, Siddhartha Chauhan, and Lalit Kumar Awasthi. “Human activity recogni- tion (har) using deep learning: Review, methodologies, progress and future research directions.” Archives of Computational Methods in Engineering 31.1 (2024): 179-219.
  24. Khan, Danyal, et al. “Advanced IoT-based human activity recognition and localization using Deep Polynomial neural network.” Ieee Access 12 (2024): 94337-94353.
  25. Yang, Xiaoshan, et al. “Cross-modal federated human activity recognition.” IEEE Transactions on Pattern Analysis and Machine Intelligence 46.8 (2024): 5345-5361.

 

IZVOR: Proceedings of the 16th International Conference on Business Information Security BISEC’2025