RGB-based Early Detection System of Drowning Person using Machine Learning Vision assisted GMM-FSM Framework for Real-time Drowning Detection
DOI:
https://doi.org/10.71426/jmt.v3.i1.pp367-379Keywords:
Drowning detection, Vision-based, Sequential modification, Hidden Markov Models (HMMs), Gaussian mixture model.Abstract
Drowning is one of the leading causes of accidental death worldwide, particularly among children and individuals in swimming environments. Despite advancements in surveillance and monitoring technologies, early detection of drowning incidents remains a major challenge due to visually complex aquatic environments, dynamic backgrounds, and swimmer interactions. Existing systems often fail to identify early behavioral symptoms or generate high false alarm rates, limiting their reliability for real-time safety applications. In this study, a novel vision-based framework is proposed for early detection of drowning incidents in swimming pools. A key challenge addressed in this work is the high noise level in foreground detection and behavior recognition caused by water reflections, background variations, and crowded scenes. To overcome these limitations, visual distress indicators and motion-based foreground descriptors are integrated to improve early identification of drowning behavior. The proposed system consists of two primary components: a vision module and an event-inference module. The vision module employs a model-based approach to accurately detect, segment, and track swimmers under varying illumination and scene conditions. The event-inference module utilizes a finite state machine to analyze swimmer motion patterns and identify abnormal behavioral transitions associated with drowning. Additionally, a sequential change detection mechanism is incorporated to enable rapid and reliable incident identification. Experimental evaluation on multiple video sequences, including simulated drowning scenarios, demonstrates that the proposed system achieves over 90% detection accuracy while maintaining low false alarm rates, confirming its effectiveness for real-time drowning prevention and aquatic safety monitoring applications.
References
[1] [Online Available]: Drowning. World Health Organization. Available from: https://www.who.int/news-room/fact-sheets/detail/drowning
[2] Liu T, He X, He L, Yuan F. A video drowning detection device based on underwater computer vision. IET Image Processing. 2023 May;17(6):1905–1918. Available from: https://doi.org/10.1049/ipr2.1276
[3] He Q, Zhang H, Mei Z, Xu X. High accuracy intelligent real-time framework for detecting infant drowning based on deep learning. Expert Systems with Applications. 2023 Oct 15;228:120204. Available from: https://doi.org/10.1016/j.eswa.2023.120204
[4] He X, Yuan F, Liu T, Zhu Y. A video system based on convolutional autoencoder for drowning detection. Neural Computing and Applications. 2023 Jul;35(21):15791–15803. Available from: https://doi.org/10.1007/s00521-023-08526-9
[5] Sakthivel V, Suriya E, Lee JW, Prakash P. AI-based smart surveillance for drowning and theft detection on beaches using drones. Drone Technology: Future Trends and Practical Applications. 2023:243–256. Available from: https://doi.org/10.1002/9781394168002.ch10
[6] Powell L, Polsley S, Casey D, Hammond T. The real-time classification of competency swimming activity through machine learning. International Journal of Aquatic Research and Education. 2023;14(1):6. Available from: https://doi.org/10.25035/ijare.14.01.06
[7] Dulhare UN, Ali MH. Underwater human detection using faster R-CNN with data augmentation. Materials Today: Proceedings. 2023;80:1940–1945. Available from: https://doi.org/10.1016/j.matpr.2021.05.653
[8] García-Alba J, Bárcena JF, Pedraz L, Fernández F, García A, Mecías M, Costas-Veigas J, Samano ML, Szpilman D. SOSeas Web App: Decision support tool to predict drowning risk using deep neural networks. Journal of Operational Oceanography. 2024;16(2):155–174. Available from: https://doi.org/10.1080/1755876X.2021.1999107
[9] Laxton V, Howard CJ, Guest D, Crundall D. Intense classification training to increase detection of drowning swimmers. Applied Cognitive Psychology. 2023;37(1):189–199. Available from: https://doi.org/10.1002/acp.4038
[10] Pushpavalli M, Abirami P, Sivagami P, Kavitha M, Bhuvaneswari C, Memala WA. Drowning prevention system for kids learning in pool using computational technology. Proceedings of IEEE ICECONF. 2023. Available from: https://ieeexplore.ieee.org/document/10083598
[11] Kulhandjian H, Ramachandran N, Kulhandjian M, D'Amours C. Human activity classification in underwater using sonar and deep learning. Proceedings of International Conference on Underwater Networks & Systems. 2019. Available from: https://doi.org/10.1145/3366486.3366509
[12] Jensen MB, Gade R, Moeslund TB. Swimming pool occupancy analysis using deep learning. Proceedings of International Workshop on Multimedia Content Analysis in Sports. 2018. Available from: https://doi.org/10.1145/3265845.3265846
[13] Chungath TT, Nambiar AM, Mittal A. Few-shot learning based underwater sonar image classification. IEEE Journal of Oceanic Engineering. 2024;49(1):294–310. Available from: https://ieeexplore.ieee.org/document/10044968
[14] Hayat MA, Yang G, Iqbal A. Mask R-CNN based real time drowning detection system. IEEE MAJICC. 2022. Available from: https://ieeexplore.ieee.org/document/9994135
[15] He Q, Mei Z, Zhang H, Xu X. Real-time infant drowning detection using YOLOv5 and Faster R-CNN. Journal of Social Computing. 2023;4(1):62–73. Available from: https://doi.org/10.23919/JSC.2023.0006
[16] Eng HL. Swimmer behavior analysis and early drowning detection. Advanced Topics in Biometrics. 2012. Available from: https://www.worldscientific.com/doi/10.1142/9789814287852_0013
[17] Gu G, Gan S, Deng J, Du Y, Qiu Z, Liu J, Liu C, Zhao J. Automated diatom detection in forensic drowning diagnosis. Applied Soft Computing. 2022;122:108885. Available from: https://doi.org/10.1016/j.asoc.2022.108885
[18] Wang F, Ai Y, Zhang W. Detection of early dangerous state in swimming pool using surveillance video. Signal, Image and Video Processing. 2022;16(1):29–37. Available from: https://doi.org/10.1007/s11760-021-01953-y
[19] Alharbi N. Exploring advance approaches for drowning detection: A review. Engineering Technology & Applied Science Research. 2024;14(4):16032–16039. Available from: https://doi.org/10.48084/etasr.7804
[20] Handalage U, Nikapotha N, Subasinghe C, Prasanga T, Thilakarthna T, Kasthurirathna D. Computer vision enabled drowning detection system. ICAC Conference. 2021. Available from: https://ieeexplore.ieee.org/document/9671126
[21] Cepeda-Pacheco JC, Domingo MC. Deep learning and 5G for child drowning prevention. Sensors. 2022;22(19):7684. Available from: https://doi.org/10.3390/s22197684
[22] Xiao H, Li Y, Xiu Y, Xia Q. Outdoor swimmers detection system using deep learning. Multimedia Systems. 2023;29(1):323–332. Available from: https://doi.org/10.1007/s00530-022-00995-7
[23] Akuthota S. Enhanced image fusion through multi-scale adaptive weighting. Journal of Computing and Data Technology. 2025;1(1):59–67. Available from: https://doi.org/10.71426/jcdt.v1.i1.pp59-67
[24] Hong D, Kim Y. Efficient swimmer detection algorithm using CNN-based SVM. Journal of Korea Society of Computer and Information. 2017;22(12):79–85. Available from: https://doi.org/10.9708/jksci.2017.22.12.079
[25] Carballo-Fazanes A, Bierens JJ. Visible behaviour of drowning persons observational study. International Journal of Environmental Research and Public Health. 2020;17(18):6930. Available from: https://doi.org/10.3390/ijerph17186930
[26] Heo JH, Yoon Y, Seo IS, Im SB, Ko YS, Lee S, Jang SJ. Evaluating sphenoid sinus fluid in drowning cases. Journal of Forensic and Legal Medicine. 2023;97:102551. Available from: https://doi.org/10.1016/j.jflm.2023.102551
[27] Zhou Y, Zhang J, Huang J. Automated diatom test in drowning cases using CNN. Forensic Science International. 2019;302:109922. Available from: https://doi.org/10.1016/j.forsciint.2019.109922
[28] Yu W, Xue Y, Knoops R. Automated diatom searching using deep neural networks. International Journal of Legal Medicine. 2021;135(2):497–508. Available from: https://doi.org/10.1007/s00414-020-02392-z
[29] Zhang J, Vieira DN. DiatomNet automatic diatom testing approach. Computer Methods and Programs in Biomedicine. 2023. Available from: https://doi.org/10.1016/j.cmpb.2023.107434
[30] Chen H, Yuan H, Qin H. Underwater drowning detection based on transformer and FPN. IEEE ICUS. 2022. Available from: https://ieeexplore.ieee.org/document/9986998
[31] Rashid AH, Razzak I, Tanveer M, Robles-Kelly A. RipNet lightweight deep neural network. Neural Information Processing Conference. 2020. Available from: https://doi.org/10.1007/978-3-030-63823-8_21
[32] Gaviria JF, Escalante-Perez A. Portable device for audio distress signal recognition. Applied Sciences. 2020;10(21):7448. Available from: https://doi.org/10.3390/app10217448
[33] Nair RS, Domnic S. Deep learning for underwater image compression. International Journal of Information Technology. 2022;14(7):3803–3814. Available from: https://doi.org/10.1007/s41870-022-01020-w
[34] Rashid AH, Razzak I, Tanveer M, Hobbs M. Improved deep architecture for rip detection. ISA Transactions. 2023;132:199–207. Available from: https://doi.org/10.1016/j.isatra.2022.05.015
[35] Khan A, Fouda MM, Do DT. Underwater target detection using deep learning. IEEE Access. 2024;12:12618–12635. Available from: https://doi.org/10.1109/ACCESS.2024.3353688
[36] Chen L, Zheng M, Duan S. Underwater target recognition based on YOLOv4. Electronics. 2021;10(14):1634. Available from: https://doi.org/10.3390/electronics10141634
[37] Hu X, Liu Y, Zhao Z. Real-time underwater detection using YOLO-V4. Computers and Electronics in Agriculture. 2021;185:106135. Available from: https://doi.org/10.1016/j.compag.2021.106135
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Muhammad Aftab Hayat, Yang Guotian, Shams ur Rehman Rehman (Author)
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
Journal of Modern Technology (JMT) Journal operates under the Attribution-NonCommercial 4.0 International (CC BY-NC 4.0). This allows others distribute, remix, tweak, and build upon the work, even commercially, as long as they credit the authors for the original creation. All authors publishing in JMT Journal accept these as the terms of publication.
