基于改进YOLOv8n的智能鱼类识别与监测系统研究与应用

Research and Application on an Intelligent System for Fish Recognition and Monitoring Based on Improved YOLOv8n

摘要: 传统海洋生物人工观测存在效率低、风险高、破坏环境等短板，难以实现长期连续监测。为此，本研究设计了一套基于改进YOLOv8n的智能鱼类识别与监测系统。通过动态加权数据增强与注意力机制优化，以及构造了适配的损失函数体系，显著提升了模型在水下复杂环境中的鲁棒性与识别精度。系统集成了算法模型、PyQt5上位机软件与边缘部署适配，创新解决了多线程任务冲突、中文标签显示失真及多源数据实时转换等工程难题。实验表明，系统在13类鱼类数据集上mAP@0.5达94%，F1分数0.904，支持毫秒级识别响应与中文可视化输出。本系统为海洋生物多样性监测、智能科考与物种预警提供了高效可靠的技术解决方案，具备重要科研与应用价值。

Abstract: The traditional manual observation of marine organisms suffers from shortcomings such as low efficiency, high risk, and environmental disruption, making it difficult to achieve long-term continuous monitoring. To address this issue, this study designs an intelligent fish recognition and monitoring system based on the improved YOLOv8n. Through dynamic weighted data augmentation, attention mechanism optimization, and the construction of an adaptive loss function system, the robustness and recognition accuracy of the model in complex underwater environments are significantly enhanced. The system integrates algorithm models, PyQt5 upper computer software, and edge deployment adaptation, innovatively solving engineering challenges including multi-thread task conflicts, Chinese label display distortion, and real-time conversion of multi-source data. Experimental results show that the system achieves a mean Average Precision (mAP@0.5) of 94% and an F1 score of 0.904 on a dataset containing 13 fish species, supporting millisecond-level recognition response and Chinese visualization output. This system provides an efficient and reliable technical solution for marine biodiversity monitoring, intelligent scientific expeditions, and species early warning, and holds significant scientific research and application value.