SENet优化的Deeplabv3+淡水鱼体语义分割*

doi:10.12398/j.issn.2096-7217.2021.01.005

智能化农业装备学报（中英文） ›› 2021, Vol. 2 ›› Issue (1): 36-43.DOI: 10.12398/j.issn.2096-7217.2021.01.005

SENet优化的Deeplabv3+淡水鱼体语义分割^*

王红君¹, 季晓宇¹, 赵辉^1,2, 岳有军¹

1.天津理工大学电气电子工程学院,天津市,300384;
2.天津农学院工程技术学院,天津市,300392

收稿日期:2021-05-01 出版日期:2021-05-15 发布日期:2021-11-30
作者简介:王红君,女,1963年生,天津人,硕士,教授;研究方向为复杂系统智能控制理论及应用。E-mail: hongewang@126.com;译者:李明心,江苏南京人,University of Illinois at Urbana-Champaign;研究方向为计算机工程。E-mail: Mingxin9@illinois.edu
基金资助:
*天津市科技支撑计划项目(17ZXYENC00080、18YFZCNC01120、15ZXZNGX00290)

SENet optimized Deeplabv3+ freshwater fish body semantic segmentation^*

Hongjun Wang¹, Xiaoyu Ji¹, Hui Zhao^1,2, Youjun Yue¹

1. College of Electrical and Electronic Engineering, Tianjin University of Technology, Tianjin, 300384, China;
2. College of Engineering and Technology, Tianjin Agricultural University, Tianjin, 300392, China

Received:2021-05-01 Online:2021-05-15 Published:2021-11-30
About author:Hongjun Wang, Professor, research direction: theory and application of complex system intelligent control. E-mail: hongewang@126.com;Translator:Mingxin Li, University of Illinois at Urbana-Champaign, research direction: computer engineering. E-mail: Mingxin9@illinois.edu
Supported by:
*Tianjin Science and Technology Support Plan Project (17ZXYENC00080、18YFZCNC01120、15ZXZNGX00290)

摘要/Abstract

摘要： 淡水鱼头、腹、鳍的各部分快速识别与精准定位是机器人实现淡水鱼快速抓取,精确切割、提升作业效率关键技术的前提。针对深度学习的淡水鱼体语义分割算法在编码特征提取阶段产生大量无效的特征通道,以及网络不断下采样和池化操作使得鱼体某些细节信息被丢失,网络性能下降、边缘分割效果不佳的问题,提出了一种基于SENet优化后的Deeplabv3+淡水鱼头、腹、鳍的语义分割算法。利用空洞/带孔卷积(dilated/atrous convolutions)实现扩展感受野,克服细节信息丢失,达到准确定位的目的,同时SENet的优化使得Deeplabv3+通过学习的方式提升淡水鱼有用的特征并抑制对当前任务用处不大的特征,最终淡水鱼各部分的语义分割平均交并比(MIoU)在自建的淡水鱼数据集上达到了93%左右,性能得到了明显提升并达到了目前先进分割水平。

关键词: 识别, 定位, 深度学习, 特征通道, 感受野, 语义分割

Abstract: The rapid identification and precise positioning of the freshwater fish's head, abdomen, and fins are prerequisites for the robot to achieve fast capturing, accurate segmentation, and the core of working efficiency improvement. The freshwater fish body's semantic segmentation algorithm conducts many invalid feature channels under the encoding feature extraction stage, and the constant network sample collection and pooling operation lead to the loss of detailed information of some fish bodies, decreasing the performance and poor the performance effect of edge cutting. A semantic segmentation algorithm base on SENet optimized Deeplabv3+ freshwater fish head, abdomen, and fins is proposed. Using dilated/atrous convolutions to enlarge the receptive field, prevent detail information loss, and actualize the precise positioning. Meanwhile, the SENet optimization enables Deeplabv3+ to improve the useful feature of freshwater fish and restrains its ineffectual feature. Eventually, the semantic segmentation MIoU of each part of the freshwater fish reached about 93% on the self-established freshwater fish data set. Its feature improved significantly and reached the current advanced segmentation level.

Key words: recognition, location, deep learning, characteristic channel, receptive field, semantic segmentation

中图分类号:

S24: TP391.4

王红君, 季晓宇, 赵辉, 岳有军. SENet优化的Deeplabv3+淡水鱼体语义分割^*[J]. 智能化农业装备学报（中英文）, 2021, 2(1): 36-43.

Hongjun Wang, Xiaoyu Ji, Hui Zhao, Youjun Yue. SENet optimized Deeplabv3+ freshwater fish body semantic segmentation^*[J]. Journal of Intelligent Agricultural Mechanization, 2021, 2(1): 36-43.

参考文献

[1] Storbeck F, Daan B. Fish species recognition using computer vision and neural network [J]. Fisheries Research, 2001, 51(1): 11-15
[2] 付中正, 何潇, 方逵, 等. 基于改进SSD网络的西兰花叶片检测研究[J]. 中国农机化学报, 2020, 41(4): 92-97.
Fu Zhongzheng, He Xiao, Fang Kui, at al. Study on the detection of broccoli leaves based on the improved SSD networks [J]. Journal of Chinese Agricultural Mechanization, 2020, 41(4): 92-97.
[3] Otsu N. A threshold selection method from gray-level histograms [J]. Systems Man & Cybernetics IEEE Transactions on, 1979, 9(1): 62-66.
[4] Canny J. A computational approach to edge detection [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1986, 8(6): 679-698.
[5] 杨杰超, 许江淳, 陆万荣, 等. 基于计算机视觉的大黄鱼体尺测算与体质量估测[J]. 中国农机化学报, 2018, 39(6): 66-70.
Yang Jiechao, Xu Jiangchun, Lu Wanrong, at al. Computer vision-based body size measurement and weight estimation of large yellow croaker [J]. Journal of Chinese Agricultural Mechanization, 2018, 39(6): 66-70.
[6] 张军文, 陈庆余, 沈建, 等. 开背大黄鱼去脏用鱼体仿形料槽设计研究[J]. 中国农机化学报, 2019, 40(7): 63-67.
Zhang Junwen, Chen Qingyu, Shen Jian, at al. Design and research of shape-like feeding tank for scratching Pseudosciaena crocea [J]. Journal of Chinese Agricultural Mechanization, 2019, 40(7): 63-67.
[7] Shi G, Hu J, Da L, et al. Dynamic artificial neural networks based on the target featuer and aplication in target recognition [C]. Conferece on Robotics and Biomimetics, New york: IEEE, 2007: 2106-2109.
[8] 王聃, 柴秀娟. 机器学习在植物病害识别研究中的应用[J]. 中国农机化学报, 2019, 40(9): 171-180.
Wang Dan, Chai Xiujuan. Application of machine learning in plant diseases recognition [J]. Journal of Chinese Agricultural Mechanization, 2019, 40(9): 171-180.
[9] Hu J, Shen L, Sun G. Squeeze-and excitation networks [C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018: 7132-7141.
[10] Chen L C, Zhu Y K, Papandreou G, at al. Encoder-decoder with atrous separable convolution for semantic image segmentation[C]. Proceedings of the European Conference on Computer Vision, 2018: 801-818.
[11] He K, Zhang X, Ren S, at al. Spatial pyramid pooling in deep convolutional networks for visual recognition [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37(9): 1904-1916.
[12] Long J,Shelhamer E, Darrell T. Fully convolutional networks for semantic segmentation [J]. IEEE Conference on Computer Vision and Pattern Recognition,2015, 5(2): 3431-3440.
[13] Zhou X, Yang G. Normalization in training U-NET for 2D biomendical semantic segmentation [J]. IEEE Robotics and Automation Letters, 2019, 4(2): 1792-1799.
[14] Chen L C, Papandreou G, Kokkinos I, et al. Semantic image segmentation with deep convolutional nets and fully connected CRFs [J]. Computer Science, 2014, 4(6): 357-361.
[15] Chen L C, Papandreou G, Kokkinos I, at al. DeepLab: semantic image segmentation with deep convolutional nets,atrous convolution,and fully connected CRFs [J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2018, 40(4): 834-848.
[16] Yurtkulu S C, Sahin Y H, Unal G. Semantic segmentation with extended deeplabv3 architecture [C]. Signal Processing and Communications Applications Conference, 2019: 1-4.
[17] 张潜, 桑军, 吴伟群, 等. 基于Xception的细粒度图像分类[J]. 重庆大学学报, 2018, 41(5): 85-91.
Zhang Qian, Sang Jun, Wu Weiqun, at al. Fine-grained image classification based on Xception [J]. Journal of Chongqing University, 2018, 41(5): 85-91.
[18] 张云, 王志伟, 史鹏坤, 等. 基于机器视觉的棉花异性纤维检测技术优化研究[J]. 中国农机化学报, 2018, 39(9): 61-65.
Zhang Yun, Wang Zhiwei, Shi Pengkun, at al. Optimization of cotton heterosexual detection technology based on machine vision [J]. Journal of Chinese Agricultural Mechanization, 2018, 39(9): 61-65.

SENet优化的Deeplabv3+淡水鱼体语义分割^*

SENet optimized Deeplabv3+ freshwater fish body semantic segmentation^*

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 8

编辑推荐

Metrics

本文评价

[1]	程振, 宋井玲, 徐洪岑, 张明亮. 基于机器视觉的蒜种智能定向技术综述[J]. 智能化农业装备学报（中英文）, 2023, 4(2): 63-70.
[2]	闫凯, 高悦, 戴百生, 孙红敏, 尹艳玲, 沈维政. 基于无锚时序动作定位的群养生猪争斗行为检测研究[J]. 智能化农业装备学报（中英文）, 2023, 4(1): 17-25.
[3]	唐瑜嵘, 沈明霞, 薛鸿翔, 陈金鑫. 人工智能技术在畜禽养殖业的发展现状与展望[J]. 智能化农业装备学报（中英文）, 2023, 4(1): 1-16.
[4]	白学峰, 常江雪, 滕兆丽, 鲁植雄. 我国智能农业拖拉机关键技术研究进展[J]. 智能化农业装备学报（中英文）, 2022, 3(2): 10-21.
[5]	张磊, 刘义亭, 陈光明, 李佩娟. 基于超声波传感器的巡检机器人导航纠偏研究[J]. 智能化农业装备学报（中英文）, 2022, 3(2): 64-70.
[6]	张隆梅, 刘岗微, 齐彦栋, 杨腾祥, 金诚谦. 农业机械无人驾驶系统关键技术研究进展[J]. 智能化农业装备学报（中英文）, 2022, 3(1): 27-36.
[7]	王国伟, 刘嘉欣. 基于卷积神经网络的玉米病害识别方法研究^*[J]. 智能化农业装备学报（中英文）, 2021, 2(1): 64-70.
[8]	王凯, 周俊, Opiyo Samwel, 张保华, 张文海. 基于三维激光雷达的履带农业机器人室外定位方法^*[J]. 智能化农业装备学报（中英文）, 2020, 1(2): 1-10.

SENet优化的Deeplabv3+淡水鱼体语义分割*

SENet optimized Deeplabv3+ freshwater fish body semantic segmentation*

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 8

编辑推荐

Metrics

本文评价

SENet优化的Deeplabv3+淡水鱼体语义分割^*

SENet optimized Deeplabv3+ freshwater fish body semantic segmentation^*