Impact of Water Velocity on Litopenaeus vannamei Behavior Using ByteTrack-Based Multi-Object Tracking

文献类型： 外文期刊

第一作者： Zhang, Jiahao

作者： Zhang, Jiahao;Zhang, Jiahao;Wang, Lei;Cui, Zhengguo;Li, Hao;Chen, Jianlei;Xu, Yong;Zhao, Haixiang;Huang, Zhenming;Qu, Keming;Cui, Hongwu;Wang, Lei;Cui, Zhengguo;Li, Hao;Chen, Jianlei;Xu, Yong;Zhao, Haixiang;Huang, Zhenming;Qu, Keming;Cui, Hongwu

作者机构：

关键词： Litopenaeus vannamei; recirculating aquaculture system; computer vision; object detection; multi-object tracking; flow velocity

期刊名称：FISHES （ 影响因子：2.4； 五年影响因子：2.4 ）

ISSN：

年卷期： 2025 年 10 卷 8 期

页码：

收录情况： SCI

摘要： In factory-controlled recirculating aquaculture systems, precise regulation of water velocity is crucial for optimizing shrimp feeding behavior and improving aquaculture efficiency. However, quantitative analysis of the impact of water velocity on shrimp behavior remains challenging. This study developed an innovative multi-objective behavioral analysis framework integrating detection, tracking, and behavioral interpretation. Specifically, the YOLOv8 model was employed for precise shrimp detection, ByteTrack with a dual-threshold matching strategy ensured continuous individual trajectory tracking in complex water environments, and Kalman filtering corrected coordinate offsets caused by water refraction. Under typical recirculating aquaculture system conditions, three water circulation rates (2.0, 5.0, and 10.0 cycles/day) were established to simulate varying flow velocities. High-frequency imaging (30 fps) was used to simultaneously record and analyze the movement trajectories of Litopenaeus vannamei during feeding and non-feeding periods, from which two-dimensional behavioral parameters-velocity and turning angle-were extracted. Key experimental results indicated that water circulation rates significantly affected shrimp movement velocity but had no significant effect on turning angle. Importantly, under only the moderate circulation rate (5.0 cycles/day), the average movement velocity during feeding was significantly lower than during non-feeding periods (p < 0.05). This finding reveals that moderate water velocity constitutes a critical hydrodynamic window for eliciting specific feeding behavior in shrimp. These results provide core parameters for an intelligent Litopenaeus vannamei feeding intensity assessment model based on spatiotemporal graph convolutional networks and offer theoretically valuable and practically applicable guidance for optimizing hydrodynamics and formulating precision feeding strategies in recirculating aquaculture systems.

分类号：

  • 相关文献
作者其他论文 更多>>

微信小程序