Abstract

Fall accidents are becoming an increasingly common and serious issue worldwide, particularly among the elderly. These incidents are not only the leading cause of injuries and fatalities in older adults but also significantly impact their quality of life. Therefore, the research and development of automatic fall detection systems have become increasingly significant. Current fall detection devices often offer wearable solutions, which can cause unnecessary inconvenience and even reduce effectiveness. Elderly individuals, especially those in poor health, may forget to wear these devices. Therefore, developing an automatic fall detection system offers a more effective solution to address this issue. The proposed system uses MediaPipe for human body key-point extraction combined with a classification model of a Long-short term memory (LSTM) network or K-Nearest-Neighbor (KNN) algorithm. It is capable of identifying the fall actions of humans in real-time environment.

References

• World Health Organization (WHO) 2021. Fact-sheets: Falls
• Rihana, S. and Mondalak, J. 2016. Wearable Fall Detection System. Middle East Conf. Biomed. Eng. (Nov. 2016), 84-87.
• Saha, B., Islam, M. S., Riad, A. K. I., Tahora S., Shahriar H. and Sneha S., 2023. BlockTheFall: Wearable Device-based Fall Detection Framework Powered by Machine Learning and Blockchain for Elderly Care. IEEE 47th Annual Computers, Software, and Applications Conference (Jun. 2023), 1412-1417.
• Amir, N. I. M., Dziyauddin, R. A., Mohamed, N., Ismail, N. S. A., Zulkifli, N. S., Din, N. M. 2022. Real-time Threshold-Based Fall Detection System Using Wearable IoT. 4th International Conference on Smart Sensors and Application (Jul. 2022), 173-178.
• Serpa, Y. R., Noqueira, M. B., Neto, P. P. M. and Rodrigues, M. A. F. 2020. Evaluating Pose Estimation as a Solution to the Fall Detection Problem. IEEE 8th International Conference on Serious Games and Applications for Health (Aug. 2020),1-7.
• Ziwei, C., Yiye, W. and Wankou, Y. 2021. Video Based Fall Detection Using Human Poses.
• Kang, Y., Kang, H. and Kim, J. 2021. Fall Detection Method Based on Pose Estimation Using GRU. Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing (Feb 2021), 169-179.
• Bugarin, C. A. Q., Lopez, J. M. M., Sambrano, M. F. C. and Loresco, P. J. M. 2022. Machine Vision-Based Fall Detection System using MediaPipe Pose with IoT Monitoring and Alarm. IEEE 10th Region 10 Humanitarian Technology Conference (Sep. 2022), 269-274
• Yuan, C., Zhang, P., Yang, Q. and Wang, J. 2022. Fall detection and Direction Judgement Based on Posture Estimation. Discrete Dynamics in Nature and Society (June 2022), 1-12.
• Guerrero, J. C. E., Espana, E. M., Anasco, M. M. and Lopera, J. E. P. 2022. Dataset for human fall recognition in an uncontrolled environment.
• Valentin, B. and Ivan, G. 2020. On-device, Real-time Body Pose Tracking with MediaPipe BlazePose.
• Asha, G. K., Jayaram, M. A. and Manjunath, A. S. 2010. Combining Akaike’s Information Criterion (AIC) and the Golden-Section Search Technique to find Optimal Numbers of K-Nearest Neighbors. International Journal of Computer Applications (May. 2010), 80-87.
• Min, W., Zou, S. and Li, J. 2018. Human fall detection using normalized shape aspect ratio. Multimedia Tools and Applications (Nov. 2018), 14331-143