«Previous Article|Table of Contents|Next Article»

Journal of Marine Science and Application[ISSN:1002-2848/CN:61-1400/f]

卷:

期数:

2026年5

页码:

1392-1408

栏目:

出版日期:

2026-11-01

Info

Title:

An Integrated Lightweight YOLO Framework Combining Structured Pruning, Knowledge Distillation, and Multi-Frame Linear Interpolation for Robust Real-Time Nonlinear Dynamic Response Analysis of Small-Scale Underwater Targets under Bubble Loading

Author(s):

Jiuqiang Wang¹;  Dongyan Shi¹;  Yuxin Gou¹;  Haifeng Zhang¹;  Xiongwei Cui²

Affilations:

Author(s):

Jiuqiang Wang¹;  Dongyan Shi¹;  Yuxin Gou¹;  Haifeng Zhang¹;  Xiongwei Cui²

1 College of Mechanical and Electrical Engineering, Harbin Engineering University, Harbin 150001, China;
2 College of Shipbuilding Engineering, Harbin Engineering University, Harbin 15001, China

Keywords:

PruneDistill-YOLOv8; Structured pruning; Knowledge distillation; Underwater target detection; Bubble dynamics

分类号:

-

DOI:

10.1007/s11804-025-00772-8

Abstract:

A You Only Look Once version 8 (YOLOv8)-based detection model optimized via pruning and knowledge distillation is proposed for the challenging task of detecting dynamic responses of small underwater bubbles. Underwater bubble experiments were conducted to collect dynamic image sequences, and the improved YOLOv8 model was used to detect bubbles and extract their nonlinear dynamic responses. Through model pruning and distillation, the optimized model considerably reduces model parameters and accelerates inference while maintaining high detection accuracy. The experimental detection results were validated by a coupled Eulerian-Lagrangian computational fluid dynamics simulation, which showed high consistency with the observed bubble dynamics. The proposed approach demonstrates excellent performance in recognizing small targets with nonlinear dynamic responses, shows strong robustness in complex underwater scenarios, and offers high inference efficiency and ease of deployment.

References:

[1] Awang MK, Aminuddin HS, Kamil NKM, Mustafa KA (2021) Class 3 wiener filtering for underwater image enhancement and restoration. 2021 International Conference on Engineering and Emerging Technologies (ICEET). Piscataway: IEEE 489-494. https://doi.org/ 10.1109/ICEET53442.2021.9659734
[2] Chaffre T, Santos PE, Le Chenadec G, Chauveau E, Sammut K, Clement B (2023) Learning adaptive control of a uuv using a bio-inspired experience replay mechanism. IEEE Access 11: 123505-123518. https://doi.org/10.1109/ACCESS.2023.3329136
[3] Cheng C, Hou X, Wang C, Wen X, Liu W, Zhang F (2024) A pruning and distillation based compression method for sonar image detection models. Journal of Marine Science and Engineering 12(6):1033. https://doi.org/10.3390/jmse12061033
[4] Cho C, Jin YS (2018) Rotary trigger double spark gap switch for repetitive pulse generator and production of silver nanocolloid. Review of Scientific Instruments 89(8): 084707. https://doi.org/10.1063/1.5035467
[5] Deng Y, Guo ZN, Huang ZG (2012) Simulation of laser-induced asymmetric bubble with lattice boltzmann method. Advanced Materials Research 591-593: 806-809. https://doi.org/10.4028/www.scientific.net/AMR.591-593.806
[6] Elmezain M, Saad Saoud L, Sultan A, Heshmat M, Seneviratne L, Hussain I (2025) Advancing underwater vision: a survey of deep learning models for underwater object recognition and tracking. IEEE Access 13: 17830-17867. https://doi.org/10.1109/ACCESS. 2025.3534098
[7] Fang S, Chi C, Liu D, Ding J, Wang K, Shi Y, Zhang C, Tang G, Chen X (2017) The development of a high speed underwater optoelectronic imaging module. Lidar Imaging Detection and Target Recognition. Bellingham: SPIE 10605106550V. https://doi.org/10.1117/12.2296358
[8] Ge S, Luo Z, Zhang C, Hua Y, Tao D (2020) Distilling channels for efficient deep tracking. IEEE Transactions on Image Processing 29: 2610-2621. https://doi.org/10.1109/TIP.2019.2950508
[9] Han W, Xing Y, Li R, Li W, Hao Y, Chen Y (2025) Dynamic characteristics of near-wall single bubble collapse-induced shock waves. Journal of Applied Fluid Mechanics 18(3): 617-630. https://doi.org/10.47176/jafm.18.3.2898
[10] Hao S, Li J, Sun X, Ma X, An B, He T (2025) MDFOaNet: A novel multi-modal pedestrian detection network based on multi-scale image dynamic feature optimization and attention mapping. IEEE Transactions on Intelligent Transportation Systems 26(1): 268-282. https://doi.org/10.1109/TITS.2024.3483892
[11] Hartmann A, Klaas M, Schroeder W (2013) Coupled airfoil heave/pitch oscillations at buffet flow. AIAA Journal 51(7): 1542-1552. https://doi.org/10.2514/1.J051512
[12] He GH, Kashiwagi M (2012) Numerical analysis of the hydroelastic behavior of a vertical plate due to solitary waves. Journal of Marine Science and Technology 17(2): 154-167. https://doi.org/10.1007/s00773-011-0155-9
[13] Ho YJ, Chang HC, Lin CW, Fan CH, Lin YC, Wei KC, Yeh CK (2021) Oscillatory behavior of microbubbles impacts efficacy of cellular drug delivery. Journal of Controlled Release 333: 316-327. https://doi.org/10.1016/j.jconrel.2021.03.044
[14] Hong Y, Liu Y, Yang S, Zhang K, Hu J (2020) Joint extraction of entities and relations using graph convolution over pruned dependency trees. Neurocomputing 411: 302-312. https://doi.org/10.1016/j.neucom.2020.06.061
[15] Hu X, Ke Y (2024) EMFANet: a lightweight network with efficient multi-scale feature aggregation for real-time semantic segmentation. Journal of Real-time Image Processing 21(2): article number 40. https://doi.org/10.1007/s11554-024-01421-z
[16] Jin X, Zhang D, Wu Q, Xiao X, Zhao P, Zheng Z (2024) Improved SiamCAR with ranking-based pruning and optimization for efficient UAV tracking. Image and Vision Computing 141: 104886. https://doi.org/10.1016/j.imavis.2023.104886
[17] Joseph KJ, Rajasegaran J, Khan S, Khan FS, Balasubramanian VN (2022) Incremental object detection via meta-learning. IEEE Transactions on Pattern Analysis and Machine Intelligence 44(12): 9209-9216. https://doi.org/10.1109/TPAMI.2021.3124133
[18] Kalateh F, Koosheh A (2018) Finite element analysis of flexible structure and cavitating nonlinear acoustic fluid interaction under shock wave loading. International Journal of Nonlinear Sciences and Numerical Simulation 19(5): 459-473. https://doi.org/10.1515/ijnsns-2016-0135
[19] Kaytoue M, Plantevit M, Zimmermann A, Bendimerad A, Robardet C (2017) Exceptional contextual subgraph mining. Machine Learning 106(8): 1171-1211. https://doi.org/10.1007/s10994-016-5598-0
[20] Kazantzidou C, Perez T, Valentinis F, Donaire A (2021) On the tuning of a nonlinear energy-based regulator for the positioning of a fully actuated surface marine craft. IET Control Theory and Applications 15(6): 850-860. https://doi.org/10.1049/cth2.12087
[21] Kedrinskii VK (2016) Multiphase systems: Wave processes, state dynamics, and mathematical models. Journal of Physics: Conference Series 754: 042007. http://dx.doi.org/10.1088/1742-6596/754/4/042007
[22] Li C, Teng X, Ding Y, Lan L (2024) Instance-level scaling and dynamic margin-alignment knowledge distillation for remote sensing image scene classification. Remote sensing 16(20): 3853. https://doi.org/10.3390/rs16203853
[23] Li JB, Xu WL, Zhai YW, Luo J, Wu H, Deng J (2022) Experimental study of the collapse strength of a bubble merged from a spark-induced bubble and a pre-existing air bubble. Ocean Engineering 249: 110907. https://doi.org/10.1016/j.oceaneng.2022.110907
[24] Li S, Saade Y, van der Meer D, Lohse D (2021) Comparison of boundary integral and volume-of-fluid methods for compressible bubble dynamics. International Journal of Multiphase Flow 145: 103834. https://doi.org/10.1016/j.ijmultiphaseflow.2021.103834
[25] Liang W, Lin Y, Zhu C, Zhang Y, Chen R (2022) Experimental study on dynamic response of aluminum film under two spark bubbles coupling loading. Ocean Engineering 266(Part 3): 112888. https://doi.org/10.1016/j.oceaneng.2022.112888
[26] Lin MG, Huang CT, Chuang YC, Chen YT, Hsu YT, Chen YK, Chou JJ, Liu TT, Shih CS, Wu AY (2022) D-NAT: Data-driven non-ideality aware training framework for fabricated computing-in-memory macros. IEEE Journal on Emerging and Selected Topics in Circuits and Systems 12(2): 381-392. https://doi.org/10.1109/JETCAS.2022.3171268
[27] Lin X, Sun Y, Liu Y, Jiang K, Cao A (2021) Stabilization of high-energy cathode materials of metal-ion batteries: control strategies and synthesis protocols. Energy & Fuels 35(9): 7511-7527. https://doi.org/10.1021/acs.energyfuels.1c00493
[28] Liu F, Xu J, Zhang L, Du R, Su Z, Zhang A, Hu Z (2022) DESNN algorithm for communication network intrusion detection. Wireless Personal Communications 126(2): 1705-1720. https://doi.org/10.1007/s11277-022-09817-5
[29] Liu H, Ye X, Zhou H, Huang H (2025) Research on UUV carrying forward-looking sonar for target location based on spatial analysis. IEEE Transactions on Instrumentation and Measurement, 74: Article No. 1002511. https://doi.org/10.1109/TIM.2025.3533664
[30] Liu Y, Wang Y, Wang H, Dou S, Tian H, Gan W, Yuan Q (2023) Dynamic template directed construction of three-dimensional porous bismuth aerogels for high-rate Na-ion storage. Journal of Materials Chemistry A 11(11): 5945-5955. https://doi.org/10.1039/d2ta09710a
[31] Liu Z, Guan X, Zhang L, Zhang Y, Pei Y, Liu C, Zhou H, Liu B, Yan K (2021) Investigations of dynamics of a single spark-induced bubble in saline water. Journal of Physics D-Applied Physicsv 54(7): 075203. https://doi.org/10.1088/1361-6463/abc4aa
[32] Lu M, Zhou B, Bu Z (2023) Two-stage distillation-aware compressed models for traffic classification. IEEE Internet of Things Journal 10(16): 14152-14166. https://doi.org/10.1109/JIOT.2023.3263487
[33] Lu Z, Chen Y, Li S, Ma M (2023) SDGC-YOLOv5: A more accurate model for small object detection. Artificial Neural Networks and Machine Learning, ICANN 2023, PT VII. Cham:Springer 14260: 199-209. https://doi.org/10.1007/978-3-031-44195-0_17
[34] Mang S, Yu Y, Zhang Q (2018) Numerical models for interactions of ablating-capillary plasma with air and bulk-loaded liquid. IEEE Transactions on Plasma Science 46(8): 3065-3074. https://doi.org/10.1109/TPS.2018.2855062
[35] Paquette Y, Pellone C, Fivel M, Ghigliotti G, Johnsen E, Franc JP (2015) Modeling fluid-structure interaction in cavitation erosion: preliminary results. Journal of Physics: Conference Series, 9th International Symposium on Cavitation (CAV2015), 656: 012053. https://doi.org、10.1088/1742-6596/656/1/012053
[36] Pietron M, Zurek D, Sniezynski B (2023) Speedup deep learning models on GPU by taking advantage of efficient unstructured pruning and bit-width reduction. Journal of Computational Science 67: 101971. https://doi.org/10.1016/j.jocs.2023.101971
[37] Ragunathan S (2016) Numerical simulation of the flow field of a confined, submerged slot jet impinging on an oscillating surface: a parametric study. Proceedings of the ASME Fluids Engineering Division Summer Meeting, West Conshohocken: ASTM 1A: Paper No: FEDSM2016-7743. https://doi.org/10.1115/FEDSM2016-7743
[38] Rozhdestvensky KV (2022) Dynamics of vapor bubble in a variable pressure field. Journal of Marine Science and Application 21(3): 83-98. https://doi.org/10.1007/s11804-022-00289-4
[39] Ruan X, Liu Y, Yuan C, Li B, Hu W, Li Y, Maybank S (2021) EDP: An efficient decomposition and pruning scheme for convolutional neural network compression. IEEE Transactions on Neural Networks and Learning Systems 32(10): 4499-4513. https://doi.org/10.1109/TNNLS.2020.3018177
[40] Schontag P, Nakath D, Roehrl S, Koeser K (2022) Towards cross domain transfer learning for underwater correspondence search. Image Analysis and Processing, ICIAP 2022, PT III Cham: Springer 13233: 461-472. https://doi.org/10.1007/978-3-031-06433-3_39
[41] Thanh HV, Thu NT, Phuong TM, Vuong ND, Jo KH, Ieee (2024) Inverted bottleneck convolution module for Yolov8. 2024 33RD International Symposium on Industrial Electronics, Ulsan: ISIE 1-6. https://doi.org/10.1109/ISIE54533.2024.10595778
[42] Tonellotto N, Macdonald C, Ounis I (2018) Efficient query processing for scalable web search. Foundations and Trends in Information Retrieval 12(4-5): 319-500. https://doi.org/10.1561/1500000057
[43] Trivedi C (2017) A review on fluid structure interaction in hydraulic turbines: A focus on hydrodynamic damping. Engineering Failure Analysis 77: 1-22. https://doi.org/10.1016/j.engfailanal.2017. 02.021
[44] Wang X, Huang F, Cheng Y (2016) Computational performance optimization of support vector machine based on support vectors. Neurocomputing 211: 66-71. https://doi.org/10.1016/j.neucom. 2016.04.059
[45] Xian Y, Yu Y, Lian Y, Fan J, Wang Z (2023) An EA-based pruning on improved YOLOv3 for rapid copper elbow surface defect detection. Engineering Applications of Artificial Intelligence, 123(Part C): 106412. https://doi.org/10.1016/j.engappai.2023.106412
[46] Xiao Y, Xiang H, Wang T, Wang Y (2024) Enhanced Industrial Action Recognition Through Self-Supervised Visual Transformers. IEEE Access 12: 134133-134143. https://doi.org/10.1109/ACCESS.2024. 3455749
[47] Xie Y, Chen T, Xu J (2023) Advancing underwater acoustic target recognition via adaptive data pruning and smoothness-inducing regularization. Arxiv arXiv: 2304. 11907. https://doi.org/10.48550/arXiv.2304.11907
[48] Zhang AM, Li SM, Xu RZ, Pei SC, Li S, Liu YL (2024) A theoretical model for compressible bubble dynamics considering phase transition and migration. Journal of Fluid Mechanics 999: A58. https://doi.org/10.1017/jfm.2024.954
[49] Zhang W, An Y (2013) Instability of a bubble chain. Physical Review E 87(5): 053023. https://doi.org/10.1103/PhysRevE.87.053023
[50] Zhao GL, Xu ZY (2025) Coastal vessel target detection model based on improved YOLOv7. Journal of Marine Science and Application. Published online. https://doi.org/10.1007/s11804-025-00635-2
[51] Zheng LN, Xuan P, Huang J, Li JL (2023) Development and application of spark-induced breakdown spectroscopy. Spectroscopy and Spectral Analysis 43(3): 665-673. https://doi.org/10.3964/j.issn.1000-0593(2023)03-0665-09
[52] Zhou J, Yang T, Zhang W (2023) Underwater vision enhancement technologies: a comprehensive review, challenges, and recent trends. Applied Intelligence 53(3): 3594-3621. https://doi.org/10.1007/s10489-022-03767-y
[53] Zhou J, Zhang B, Zhang D, Vivone G, Jiang Q (2024) DTKD-Net: Dual-teacher knowledge distillation lightweight network for water-related optics image enhancement. IEEE Transactions on Geoscience and Remote Sensing 62: Article No. 4207213. https://doi.org/10.1109/TGRS.2024.3422667

Memo

Memo:

Received date:2025-4-2;Accepted date:2025-6-3。<br>Foundation item:This study is supported by the National Natural Science Foundation of China (12172100), the National Natural Science Foundation of China (52201359), and the Doctoral Research Innovation Fund (3072022GIP0701).<br>Corresponding author:Xiongwei Cui,E-mail:cuixiongwei@hrbeu.edu.cn

Commonly used function

Last Update: 2026-11-04