Recognition of Acetes chinensis fishing vessel based on 3-2D integration model behavior
-
摘要: 针对中国毛虾 (Acetes chinensis) 产量逐年锐减问题,中国开始对近海海域实施毛虾限额捕捞措施,采用视频监控技术辅助捕捞管理。提出一种基于3-2D融和模型的毛虾捕捞渔船行为识别方法,为限额捕捞管理提供新的解决方案。通过在毛虾渔船上4个固定位置安装高清摄像设备,并记录捕捞作业全过程,共获取600余个视频监控数据作为初始数据;从初始数据中筛选有效的视频数据,同时对视频数据进行5种行为的划分和标记。为了提高网络训练的效率,对视频数据进行压缩和帧数分割等预处理;最后,通过搭建3-2D融合的卷积神经网络来训练模型,实现渔船行为特征的提取和分类。结果表明,捕捞渔船行为识别方法的分类精度为95.35%,召回率为94.50%,平均精确度为96.60%,模型整体得分达93.32%,平均检测时间为35.46 ms·帧−1,可用于毛虾渔船捕捞视频的实时分析。Abstract: Since the yield of Acetes chinensis has decreased sharply year by year, China has begun to implement quota fishing measures for A. chinensis in offshore waters by using video surveillance technology to assist the fishing management. This paper proposes a method for identifying the behavior of A. chinensis fishing vessels based on the 3-2D fusion model, so as to provide a new solution for quota fishing management. By installing high-definition camera equipment at four fixed positions on the A. chinensis fishing vessel and recording the entire process of fishing operations, we had obtained more than 600 video surveillance data had been as initial data. Secondly, we filtered effective video data from the initial data, and divided and labeled them with five behaviors. In order to improve the efficiency of network training, we preprocessed the video data such as compression and frame number segmentation. Finally, the model was trained by building a 3-2D fusion convolutional neural network to realize the extraction and classification of fishing vessel behavior characteristics. The results show that the classification accuracy of the fishing vessel behavior recognition method was 95.35%; the recall rate was 94.50%; the average accuracy was 96.60%; the overall score of the model could reach 93.32%; and the average detection time was 35.46 ms·frame−1. The method can be used for real-time analysis of the fishing video of A. chinensis fishing boats.
-
Key words:
- Acetes chinensis /
- Quota fishing /
- Deep learning /
- Convolutional neural network
-
表 1 选用视频数据
Table 1. Selection of video data
渔船标签
Fishing vessel label渔船行为
Fishing vessel behavior渔船视频数量
Number of fishing vessel videos/个0 停靠 Dock 90 1 航行 Sail 110 2 下网 Cast net 110 3 收网 Put away net 110 4 等待 Wait 80 表 2 制作数据标签
Table 2. Production of data labels
视频路径
Video path视频帧数
Video frames捕捞渔船标签
Fishing vessel label行为状态
Behavioral statesyy/d01/D01_20210617171455.mp4 0~13 230 0 停靠 Dock syy/d01/D01_20210620043651.mp4 0~11 940 1 航行 Sail syy/d01/D01_20210620043651.mp4 11 940~50 093 2 下网 Cast net syy/d01/D01_20210630181131.mp4 0~69 301 3 收网 Put away net syy/d01/D01_20210620061938.mp4 0~4 330 4 等待 Wait 表 3 模型评价主要指标及结果
Table 3. Main indicators and results of model evaluation
指标
Index训练结果
Training result测试结果
Test result精度 Precision/% 99.60 95.35 召回率 Recall rate/% 99.63 94.50 平衡F分数 F1-Score/% 99.06 93.32 平均精确度 AP/% 98.70 96.60 时间 t/(ms·帧−1) — 35.46 -
[1] 李国东, 仲霞铭, 熊瑛, 等. 基于北斗船位数据的渔业信息解译与应用研究——以中国毛虾限额捕捞管理为例[J]. 海洋与湖沼, 2021, 52(3): 746-753. doi: 10.11693/hyhz20201000288 [2] 农业农村部渔业渔政管理局, 全国水产技术推广总站, 中国水产学会. 2019年中国渔业统计年鉴[J]. 北京: 中国农业出版社, 2020: 2. [3] MULYANI S, VESTIYATI P M, ALAMSYAH H K, et al. Effect of differences in salt concentration on the quality of rebon shrimp paste (Acetes sp. ) in Tegal district[J]. IOP Conf Ser:Earth Environ Sci, 2021, 755(1): 012051. doi: 10.1088/1755-1315/755/1/012051 [4] UNG T D, TRAN T Y N, NGUYEN N L, et al. Biochemical characteristics of small shrimp (Acetes japonicus) of varied sizes collected in Ben Tre Province, Vietnam[J]. IOP Conf Ser: Mater Sci Eng, 2021, 1092(1): 012080. doi: 10.1088/1757-899X/1092/1/012080 [5] 刘洋, 张胜茂, 王斐, 等. 海洋捕捞鱼类BigH神经网络分类模型设计与实现[J]. 工业控制计算机, 2021, 34(6): 18-20. doi: 10.3969/j.issn.1001-182X.2021.06.007 [6] 梁建胜, 温贺平. 基于深度学习的视频关键帧提取与视频检索[J]. 控制工程, 2019, 26(5): 965-970. [7] 蒋梦迪, 程江华, 陈明辉, 等. 视频和图像文本提取方法综述[J]. 计算机科学, 2017, 44(S2): 8-18. [8] 李秀智, 张冉, 贾松敏. 面向助老行为识别的三维卷积神经网络设计[J]. 北京工业大学学报, 2021, 47(6): 589-597. [9] 耿家利. 渔船轨迹大数据存储优化与行为识别技术研究[D]. 杭州: 杭州电子科技大学, 2018: 5-8 [10] ZHANG H, YANG S L, FAN W, et al. Spatial analysis of the fishing behaviour of tuna purse seiners in the western and central Pacific based on vessel trajectory date[J]. J Mar Sci Eng, 2021, 9(3): 322. doi: 10.3390/jmse9030322 [11] 宁耀. 基于深度学习的渔船行为识别方法研究[D]. 兰州: 兰州大学, 2020: 42-53. [12] TRAN D, BOURDEV L, FERGUS R, et al. Learning spatiotemporal features with 3D convolutional networks[C]. 2015 Proceedings of the IEEE International Conference on Computer Vision (ICCV), Santiago, Chile. New York: IEEE, 2015: 4489-4497. [13] CARREIRA J, ZISSERMAN A. Quo vadis, action recognition? A new model and the kinetics dataset[C]. 2017 Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu HI, USA. New York: IEEE, 2017: 6299-6308. [14] 王书献, 张胜茂, 朱文斌, 等. 基于深度学习YOLOV5网络模型的金枪鱼延绳钓电子监控系统目标检测应用[J]. 大连海洋大学学报, 2021, 36(5): 1-17. [15] 王书献, 孙永文, 张胜茂, 等. 基于卫星AIS远洋船位的热力图自动制图[J]. 渔业信息与战略, 2021, 36(1): 45-53. [16] 裴凯洋, 张胜茂, 樊伟, 等. 浙江省帆张网捕捞强度分布的提取方法[J]. 水产学报, 2020, 44(11): 1913-1925. [17] 张胜茂, 樊伟, 张衡, 等. 远洋捕捞渔船电子监控视频文字信息提取[J]. 渔业信息与战略, 2020, 35(2): 141-146. [18] 汤先峰, 张胜茂, 樊伟, 等. 基于深度学习的刺网与拖网作业类型识别研究[J]. 海洋渔业, 2020, 42(2): 233-244. doi: 10.3969/j.issn.1004-2490.2020.02.011 [19] 王书献, 张胜茂, 戴阳, 等. 利用声呐数据提取磷虾捕捞深度方法研究[J]. 南方水产科学, 2021, 17(4): 91-97. doi: 10.12131/20210020 [20] 农业农村部印发《通告》规定2021年伏休期间特殊经济品种专项捕捞许可[J]. 中国水产, 2021(6): 6-8. [21] IOFFE S, SZEGEDY C. Batch normalization: accelerating deep network training by reducing internal covariate shift[C]. 2015 International conference on machine learning (ICML), Lille, France. San Diego, CA: JMLR-Journal Machine Learning Research, 2015: 448-456. [22] HE K, ZHANG X, REN S, et al. Delving deep into rectifiers: surpassing human-level performance on imagenet classification[C]. 2015 Proceedings of the IEEE International Conference on Computer Vision (ICCV), Santiago, Chile. New York: IEEE, 2015: 1026-1034. [23] 崔雪森, 田晓清, 康伟, 等. 基于卷积神经网络的微藻种类识别[J]. 上海海洋大学学报, 2021, 30(4): 710-717. [24] 罗晨辉, 孙洪飞. 改进型的batch normalization: BNalpha[J]. 计算机应用研究, 2021, 38(6): 1870-1873. [25] WEI Q J, WANG W B. Research on image retrieval using deep convolutional neural network combining L1 regularization and PRelu activation function[J]. IOP Conf Ser: Earth Environ Sci, 2017, 69(1): 012156. [26] 张胜茂, 刘洋, 樊伟, 等. 基于TensorFlow的水族馆鱼类目标检测APP开发[J]. 渔业现代化, 2020, 47(2): 60-67. doi: 10.3969/j.issn.1007-9580.2020.02.008 [27] 张超. P−R曲线与模型评估问题研究[J]. 现代信息科技, 2020, 4(4): 23-24. doi: 10.3969/j.issn.2096-4706.2020.04.007 [28] 陈冠宇. 基于深度学习的小目标检测方法研究[D]. 武汉: 中国地质大学, 2020: 27-29. [29] 马金涛. 基于深度学习的复杂环境目标检测方法研究[D]. 哈尔滨: 哈尔滨理工大学, 2021: 27-28. [30] 李琪. 基于深度学习的小目标检测方法研究[D]. 成都: 中国科学院大学 (中国科学院光电技术研究所), 2021: 13-14. [31] 赵亮, 胡杰, 刘汉, 等. 基于语义分割的深度学习激光点云3D目标检测算法[J]. 中国激光: 2021, 48(17): 171-183. [32] 李新德, 刘苗苗, 徐叶帆, 等. 一种基于2D和3D SIFT特征级融合的一般物体识别算法[J]. 电子学报, 2015, 43(11): 2277-2283. doi: 10.3969/j.issn.0372-2112.2015.11.021 [33] 卢元兵, 李华朋, 张树清. 基于混合3D-2D CNN的多时相遥感农作物分类[J]. 农业工程学报, 2021, 37(13): 142-151. doi: 10.11975/j.issn.1002-6819.2021.13.017 [34] 黄海广, 胡乃军, 仇志金, 等. 基于多源定位数据的实时船舶监控系统设计[J]. 中国海洋大学学报(自然科学版), 2015, 45(9): 122-129. [35] KROODSMA D A, MAYORGA J, HOCHBERG T, et al. Tracking the global footprint of fisheries[J]. Science, 2018, 359(6378): 904. doi: 10.1126/science.aao5646 [36] GUAN Y N, ZHANG J, ZHANG X, et al. Identification of fishing vessel types and analysis of seasonal activities in the northern South China Sea based on AIS Data: a case study of 2018[J]. Remote Sens, 2021, 13(10): 1952. doi: 10.3390/rs13101952 [37] 裴凯洋, 张胜茂, 樊伟, 等. 基于VMS的张网渔船捕捞努力量与网位坐标提取方法[J]. 上海海洋大学学报, 2021, 30(1): 179-188. [38] 朱浩朋, 伍玉梅, 唐峰华, 等. 采用卷积神经网络构建西北太平洋柔鱼渔场预报模型[J]. 农业工程学报, 2020, 36(24): 153-160,57. doi: 10.11975/j.issn.1002-6819.2020.24.018 [39] 范秀梅, 杨胜龙, 张胜茂, 等. 基于栖息地指数的阿拉伯海鲐鱼渔情预报模型构建[J]. 南方水产科学, 2020, 16(4): 8-17. doi: 10.12131/20190255 [40] 郑巧玲, 樊伟, 张胜茂, 等. 基于神经网络和VMS的渔船捕捞类型辨别[J]. 南方水产科学, 2016, 12(2): 81-87. doi: 10.3969/j.issn.2095-0780.2016.02.012 -