IRO Journals
Journal of Trends in Computer Science and Smart Technology

Efficient Skeleton-Based Human Action Recognition for IoT Edge Devices Using Lightweight LSTM Networks

Open Access
Published: 16 July, 2026
Pages: 649-664
  • Shubbar Salman Baqer Alkhafaji , Shubbar Salman Baqer Alkhafaji
    Department of Computer Science, Faculty of Education, University of Kufa, Najaf, Iraq
    Department of Computer Science, Faculty of Education, University of Kufa, Najaf, Iraq
  • Ali Abdulazeez Mohammed Baqer Qazzaz , Ali Abdulazeez Mohammed Baqer Qazzaz
    Department of Computer Science, Faculty of Education, University of Kufa, Najaf, Iraq
    Department of Computer Science, Faculty of Education, University of Kufa, Najaf, Iraq
  • Yousif Samer Mudhafar Yousif Samer Mudhafar
    Department of Computer Science, Faculty of Education, University of Kufa, Najaf, Iraq
    Department of Computer Science, Faculty of Education, University of Kufa, Najaf, Iraq
view PDF
view PDF

How to Cite

Alkhafaji, Shubbar Salman Baqer, Ali Abdulazeez Mohammed Baqer Qazzaz, and Yousif Samer Mudhafar. 2026. “Efficient Skeleton-Based Human Action Recognition for IoT Edge Devices Using Lightweight LSTM Networks”. Journal of Trends in Computer Science and Smart Technology 8 (3): 649-64. https://doi.org/10.36548/jtcsst.2026.3.011.
ACM ACS APA ABNT Chicago Harvard IEEE MLA Turabian Vancouver AMA

Download Citation

Endnote/Zotero/Mendeley (RIS) BibTeX

Keywords

Human Action Recognition (HAR)
IoT
Skeleton-based Classification
Long Short-Term Memory (LSTM)
MediaPipe
Edge Computing
Skeletal Representation

Abstract

Human Action Recognition (HAR) is becoming increasingly important in areas such as intelligent surveillance, healthcare monitoring, and facilitating human-computer interaction. However, most current techniques use RGB-based deep learning, which is computationally intensive and cannot run on low-resource edge devices. This study presents a lightweight, skeleton-based Human Action Recognition (HAR) framework. The framework utilizes MediaPipe pose estimation, spatiotemporal normalization, and LSTM-based temporal modeling to efficiently recognize kinetically distinct human movements on resource-limited Internet of Things (IoT) edge devices. It combines MediaPipe pose estimation with Long Short-Term Memory (LSTM) networks to model temporal motion efficiently in resource-constrained IoT edge computing environments. Twelve kinetically unique actions from the UCF50 dataset were rendered into skeletal representations by temporally interpolating and spatially normalizing the action sequences. The HAR framework achieved 85.14% classification accuracy, indicating that sufficient motion-related information can be inferred from skeletal topology, allowing for reliable recognition at a reduced computational cost. The results demonstrate a practical balance between efficiency and performance, positioning the framework as an effective tool for edge-oriented HAR applications. From an IoT perspective, the framework has the potential to support IoT-enabled smart surveillance, assisted living, and intelligent monitoring systems by making it suitable for connected cameras and edge devices for local action recognition, even with limited computing power and communication bandwidth. This makes it a promising candidate for deployment across diverse real-world environments while preserving privacy through skeleton-based representations.

References

  1. Alotaibi, Faiz, Mrim M. Alnfiai, Fahd N. Al-Wesabi, Mesfer Alduhayyem, Anwer Mustafa Hilal, and Manar Ahmed Hamza. "Optimal Deep Recurrent Neural Networks for Iot-Enabled Human Activity Recognition in Elderly and Disabled Persons." Journal of Disability Research 2, no. 2 (2023): 79-88.
  2. Martin, Pierre-Etienne, Jenny Benois-Pineau, Renaud Péteri, Akka Zemmari, and Julien Morlier. "3D Convolutional Networks for Action Recognition: Application to Sport Gesture Recognition." In Multi-Faceted Deep Learning: Models and Data, Cham: Springer International Publishing, 2012, 199-229.
  3. Eleeasa, Kevina Nugraha, Salomo Hendrian Sudjonoa, and Alexander Agung Santoso Gunawanb. "Video Based Physical Bullying Detection with Deep Learning: Utilizing 3 Dimensional and Dynamic Vision Sensors." Procedia Computer Science 269 (2025): 1077-1086.
  4. Peng, Zhiyong, Jiang Du, and Yulong Qiao. "Design of GPU Network-On-Chip for Real-Time Video Super-Resolution Reconstruction." Micromachines 14, no. 5 (2023): 1055.
  5. Spiller, E., & Giampaolo, N. (2022). Social, ethical, legal, privacy issues identification and monitoring (HAIKU Project Deliverable D1.3).
  6. Wang, Cailing, and Jingjing Yan. "A Comprehensive Survey of Rgb-Based and Skeleton-Based Human Action Recognition." IEEE Access 11 (2023): 53880-53898.
  7. Khedher, Issam, Jean-Marie Favreau, Serge Miguet, and Gilles Gesquière. "RGB2LST: Enhancing Deep Learning-Based Land Surface Temperature Estimation with Multi-Modality and Artifacts Removal." In 2024 32nd European Signal Processing Conference (EUSIPCO), IEEE, 2024, 2002-2006.
  8. O'Donnell, Kyle, and Chandra Kambhamettu. "Feature Matching in the Dark: Homography-Based RGB-IR Feature Transformation for Low-Light Vision." In Proceedings of the Computer Vision and Pattern Recognition Conference, 2025, 1703-1711.
  9. Yang, Haoxin, Weihong Chen, Xuemiao Xu, Cheng Xu, Peng Xiao, Cuifeng Sun, Shaoyu Huang, and Shengfeng He. "Starpose: 3D Human Pose Estimation via Spatial-Temporal Autoregressive Diffusion." IEEE Transactions on Circuits and Systems for Video Technology (2025).
  10. Pan, Xiaojian, Guang Li, Ningfei Zhang, and Jianjun Li. "3D Human Pose Estimation Based on a Hybrid Approach of Transformer and GCN-Former." Journal of Visual Communication and Image Representation (2025): 104696.
  11. Jiang, Jia-Hong, and Nan Xia. "Pcnet: A Human Pose Compensation Network Based on Incremental Learning for Sports Actions Estimation." Complex & Intelligent Systems 11, no. 1 (2025): 17.
  12. Chen, Kaixin, Lin Zhang, Zhong Wang, Shengjie Zhao, and Yicong Zhou. "Skeleton-Aware Graph-Based Adversarial Networks for Human Pose Estimation from Sparse Imus." ACM Transactions on Multimedia Computing, Communications and Applications 21, no. 4 (2025): 1-22.
  13. Ali, Khalid, Zineddine Bettouche, Andreas Kassler, and Andreas Fischer. "Enhancing Spatiotemporal Networks with Xlstm: A Scalar LSTM Approach for Cellular Traffic Forecasting." In 2025 16th International Conference on Network of the Future (NoF), IEEE, 2025, 167-175.
  14. Selvaperumal, P., F. Sheeja Mary, T. L. Roy, Yousef A. Baker El-Ebiary, Gowrisankar Kalakoti, and Sandeep Kumar Mathariya. "Explainable Deep Temporal Modeling for Stroke Risk Assessment Using Attention-Based LSTM Networks." International Journal of Advanced Computer Science & Applications 16, no. 6 (2025).
  15. Chakri, Sana, and Naoual Mouhni. "Time-Series Neural Network Predictions Using LSTM for Clustering and Forecasting GPS Data of Bird Immigration." In 2024 International Conference on Global Aeronautical Engineering and Satellite Technology (GAST), IEEE, 2024, 1-4.
  16. Xiong, Xin, Weidong Min, Qing Han, Qi Wang, and Cheng Zha. "Action Recognition Using Action Sequences Optimization and Two‐Stream 3D Dilated Neural Network." Computational Intelligence and Neuroscience 2022, no. 1 (2022): 6608448.
  17. Liu, Zhen, Qin Cheng, Chengqun Song, and Jun Cheng. "Cross-Scale Cascade Transformer for Multimodal Human Action Recognition." Pattern Recognition Letters 168 (2023): 17-23.
  18. Doan, Thanh-Nghi. "An Efficient Patient Activity Recognition Using LSTM Network and High-Fidelity Body Pose Tracking." International Journal of Advanced Computer Science and Applications 13, no. 8 (2022). 226–232.
  19. Song, Yi-Fan, Zhang Zhang, Caifeng Shan, and Liang Wang. "Constructing Stronger and Faster Baselines For Skeleton-Based Action Recognition." IEEE transactions on pattern analysis and machine intelligence 45, no. 2 (2022): 1474-1488.
  20. Zhong, Enmin, Carlos R. Del-Blanco, Daniel Berjón, Fernando Jaureguizar, and Narciso García. "Real-Time Monocular Skeleton-Based Hand Gesture Recognition Using 3D-Jointsformer." Sensors 23, no. 16 (2023): 7066.
  21. Wang, Chuanchuan, Ahmad Sufril Azlan Mohmamed, Mohd Halim Bin Mohd Noor, Xiao Yang, Feifan Yi, and Xiang Li. "ARN-LSTM: A Multi-Stream Fusion Model for Skeleton-Based Action Recognition." arXiv preprint arXiv:2411.01769 (2024).
  22. Wazwaz, Ayman, Khalid Amin, Noura Semary, and Tamer Ghanem. "Light Weight Human Activity Recognition using Raspberry PI IoT Edge and Reduced Features from Smartphones." IJCI. International Journal of Computers and Information 10, no. 3 (2023): 1-8.
  23. Voillemin, Theo, Hazem Wannous, and Jean-Philippe Vandeborre. "2D Deep Video Capsule Network with Temporal Shift for Action Recognition." In 2020 25th International Conference on Pattern Recognition (ICPR), IEEE, 2021, 3513-3519.
  24. Jandhyam, Lakshmi Alekhya, Ragupathy Rengaswamy, and Narayana Satyala. "An Optimized Deep LSTM Model for Human Action Recognition." Revue d'Intelligence Artificielle 38, no. 1 (2024): 11-23.
  25. Al-muqarm, Abbas M. Ali, Yousif Mudhafar, Aboothar Mahmood Shakir, Murtada Kazem, Ruqaya Abdel-Yahiya, and Baqer Saad Abdul Zahra. "Low-Cost Smart Learning with Moodle-Based Raspberry Pi 4 for University Students." In 2023 6th International Conference on Engineering Technology and its Applications (IICETA), IEEE, 2023, 603-608.
  26. Abdulnabi, Saif Hasan, Yousif Samer Mudhafar, Ali Abdulhassan Kadhim, Mohammad Baqer Mahdi, and Hassan Hadi Sojar. "Neural Network-Based System Identification: A Comprehensive FPGA Design and Implementation." In 2024 IEEE International Conference on Artificial Intelligence and Mechatronics Systems (AIMS), IEEE, 2024, 1-7.