基于SENet的工厂化循环水养殖鳗鲡(Anguilla)数量评估研究  

ESTIMATION OF EEL(ANGUILLA)QUANTITY FARMED IN INDUSTRIAL RECIRCULATING AQUACULTURE SYSTEM BASED ON SENET

作　　者：林茜 江兴龙[1,2] 周世豪 LIN Xi;JIANG Xing-Long;ZHOU Shi-Hao(Fisheries College,Jimei University,Xiamen 361021,China;Engineering Research Center of the Modern Technology for Eel Industry,Ministry of Education,Xiamen 361021,China)

机构地区：[1]集美大学水产学院,福建厦门361021 [2]鳗鲡现代产业技术教育部工程研究中心,福建厦门361021

出　　处：《海洋与湖沼》2025年第1期206-213,共8页Oceanologia Et Limnologia Sinica

基　　金：国家重点研发计划“特色鱼类精准高效养殖关键技术集成与示范”,2020YFD0900102号;鳗鲡现代产业技术教育部工程研究中心开放基金,RE202304号,RE202101号。

摘　　要：为探索应用计算机听觉技术实现对工厂化循环水养殖鳗鲡数量的评估,建立了一种基于回归分析的SENet网络模型。针对数据集中包含的白噪声声谱图数据缺乏可利用的动态规律问题,通过修改SENet输出层、输出范围、评价指标等,使其直接利用图像进行回归分析任务,从而进一步提高了网络在图像分析任务上的性能。在循环水养殖鳗鲡的数量评估试验中,设置8组不同的鳗鲡数量进行试验,结果表明:水听器接收到的声音信号与鱼数量呈现出明显的相关性;在测试阶段SENet网络的拟合相关系数为0.98,SENet回归分析模型在测试集样本上的决定系数(R^(2))、平均绝对误差(MAE)和平均绝对百分比误差(MAPE)分别为0.96、1.66和4.43%;采集了30组数据样本对训练好的模型进行验证试验,SENet模型预测数量的相对误差与变异系数都在8%以内,预测准确率达到90%以上。A SENet network model was established based on regression analysis to estimate the quantity of eels cultured in recirculating aquaculture system(RAS).To solve the problem that the white noise spectrogram data contained in the dataset lacks available dynamic rules,the SENet output layer,output range,evaluation index,and so on were modified,by which SENet is able to use images directly for regression analysis tasks,thus further improving the performance of the network in image analysis tasks.In the estimation work,eight groups of different eel numbers were set up.Results show obvious correlation between the acoustic signal received by hydrophone and the fish quantity.The fitting correlation coefficient of SENet network reached 0.98.The determination coefficient,mean absolute error,and mean absolute percentage error of SENet regression analysis model on test set samples were 0.96,1.66,and 4.43%,respectively.Thirty sets of data samples were collected for validation tests on the trained model.The relative error and the coefficient of variation of the fish population estimation were both within 8%,and the estimation accuracy was over 90%.

关 键 词：计算机听觉技术 鳗鲡 鱼群数量评估 声音信号 SENet网络 

分 类 号：Q175[生物学—水生生物学] S931[生物学—普通生物学] S965[农业科学—渔业资源]