基于改进的DeepLabV3+网络模型的杂交水稻育种父母本语义分割研究

doi:10.3785/j.issn.1008-9209.2022.09.051

浙江大学学报（农业与生命科学版）

2023, Vol. 49

Issue (6): 893-902 DOI: 10.3785/j.issn.1008-9209.2022.09.051

农业工程

基于改进的DeepLabV3+网络模型的杂交水稻育种父母本语义分割研究

温佳¹(

),梁喜凤¹(

),王永维²

^1.中国计量大学机电工程学院，浙江杭州 310018
^2.浙江大学生物系统工程与食品科学学院，浙江杭州 310058

Research on semantic segmentation of parents in hybrid rice breeding based on improved DeepLabV3+ network model

Jia WEN¹(

),Xifeng LIANG¹(

),Yongwei WANG²

^1.College of Mechanical and Electrical Engineering, China Jiliang University, Hangzhou 310018, Zhejiang, China
^2.College of Biosystems Engineering and Food Science, Zhejiang University, Hangzhou 310058, Zhejiang, China

全文: PDF(6911 KB) HTML

摘要：

为解决杂交水稻育种授粉过程中父母本区分的精确性和实时性问题，本研究提出一种基于全卷积神经网络的、改进的DeepLabV3+杂交水稻育种父母本区分的语义分割模型。采用轻量化的主干网络MobileNetV2结构替换原DeepLabV3+的主干网络Xception结构，使之更适用于移动设备，并提出一种联系较为紧密的低层特征信息提取方法，将较低层次信息和较高层次信息初步融合作为原低层次信息的输入，使网络获得更加密集的信息，从而增强网络对于细节的提取能力。结果表明，改进的DeepLabV3+网络模型较原DeepLabV3+网络模型具有更高的杂交水稻制种父母本分割精度，并能够减少模型训练和图片预测时间。将改进后的DeepLabV3+网络模型与其他主流网络和先进网络模型对比发现，各项参数精度均有所提高。本研究为深度学习在农业视觉机器人领域中的发展提供了参考。

关键词： 语义分割; 深度学习; DeepLabV3+网络模型; 杂交水稻; 轻量化模型

Abstract:

In order to solve the precision and real-time problems of parental discrimination in the processes of hybrid rice breeding and pollination, an improved DeepLabV3+ hybrid rice breeding parental discrimination semantic segmentation model based on a fully convolution neural network was proposed. The lightweight MobileNetV2 structure of the backbone network was used to replace the Xception structure of the original DeepLabV3+ backbone network, which is more suitable for the application on mobile devices. An extraction method of low-level features with close connection was proposed. The lower-level information and higher-level information were preliminarily concated as the input of the original lower-level information, which enabled the network to obtain more intensive information, thus enhancing the ability of the network to extract details. The results showed that the improved DeepLabV3+ network model had higher segmentation precision for parents of hybrid rice seed production than the original DeepLabV3+ network model, and reduced the model training time and image predictive time. Compared with other mainstream network models and advanced network models, it is found that the accuracy of different parameters of improved DeepLabV3+ network model is improved. This study provides a reference for the development of deep learning in the field of agricultural visual robots.

Key words: semantic segmentation deep learning DeepLabV3+ network model hybrid rice lightweight model

收稿日期: 2022-09-05 出版日期: 2023-12-25

CLC:

TP391.4

基金资助: 国家自然科学基金项目(31971796)

通讯作者: 梁喜凤 E-mail: wenjia.cjlu@qq.com;lxfcjlu@163.com

作者简介: 温佳（https://orcid.org/0000-0002-0076-9798），E-mail：wenjia.cjlu@qq.com

	服务
	把本文推荐给朋友
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	温佳
	梁喜凤
	王永维

引用本文:

温佳,梁喜凤,王永维. 基于改进的DeepLabV3+网络模型的杂交水稻育种父母本语义分割研究[J]. 浙江大学学报（农业与生命科学版）, 2023, 49(6): 893-902.

Jia WEN,Xifeng LIANG,Yongwei WANG. Research on semantic segmentation of parents in hybrid rice breeding based on improved DeepLabV3+ network model. Journal of Zhejiang University (Agriculture and Life Sciences), 2023, 49(6): 893-902.

链接本文:

https://www.zjujournals.com/agr/CN/10.3785/j.issn.1008-9209.2022.09.051 或 https://www.zjujournals.com/agr/CN/Y2023/V49/I6/893

图1 晴天和阴天的水稻样本处理图像A.样本1（阴天）；B.样本1垂直翻转（阴天）；C.样本1水平翻转（阴天）；D.样本2（晴天）；E.样本2垂直翻转（晴天）；F.样本2水平翻转（晴天）；G.样本3（阴天）；H.样本3垂直翻转（阴天）；I.样本3水平翻转（阴天）。

图2 使用Labelme软件进行图像标注

图3 改进的DeepLabV3+网络模型结构图

图4 MobileNetV2网络模块的倒残差结构

表1 MobileNetV2网络模块整体操作步骤

表2 DeepLabV3+网络模型训练参数设置

图5 改进的DeepLabV3+网络模型损失曲线

图6 改进的DeepLabV3+网络模型平均交并比曲线

图7 改进的DeepLabV3+网络模型（A）与原DeepLabV3+网络模型（B）损失对比

图8 改进的DeepLabV3+网络模型与原DeepLabV3+网络模型的平均交并比对比

图9 改进的DeepLabV3+网络模型与原DeepLabV3+网络模型的多项参数对比

图10 改进的DeepLabV3+网络模型与原DeepLabV3+网络模型的分割效果图像对比

表3 不同网络模型参数对比结果

1	陈科尹,吴崇友,关卓怀,等.基于统计直方图k-means聚类的水稻冠层图像分割[J].江苏农业学报,2021,37(6):1425-1435. DOI:10.3969/j.issn.1000-4440.2021.06.009 CHEN K Y, WU C Y, GUAN Z H, et al. Rice canopy image segmentation based on statistical histogram k-means clustering[J]. Jiangsu Journal of Agricultural Sciences, 2021, 37(6): 1425-1435. (in Chinese with English abstract) doi: 10.3969/j.issn.1000-4440.2021.06.009
2	黄琼.基于图像的水稻种子品种分类识别方法研究[D].江西,南昌:江西农业大学,2022. HUANG Q. Research on image-based classification and identification of rice seed varieties[D]. Nanchang, Jiangxi: Jiangxi Agricultural University, 2022. (in Chinese with English abstract)
3	朱盼盼,张正华,郭丽瑞,等.基于ExG因子的水稻病斑分割技术[J].信息与电脑,2022,34(9):73-76. ZHU P P, ZHANG Z H, GUO L R, et al. ExG factor-based spot segmentation technology for rice[J]. China Computer & Communication, 2022, 34(9): 73-76. (in Chinese with English abstract)
4	高婷,宋少忠,张佳环,等.基于阈值分割算法的水稻叶片病斑图像分割[J].吉林省教育学院学报,2021,37(10):183-186. DOI:10.16083/j.cnki.1671-1580.2021.10.044 GAO T, SONG S Z, ZHANG J H, et al. Segmentation of the scabin rice leaf image based on threshold segmentation algorithm[J]. Journal of Educational Institute of Jilin Province, 2021, 37(10): 183-186. (in Chinese with English abstract) doi: 10.16083/j.cnki.1671-1580.2021.10.044
5	唐守锋,翟少奇,仝光明,等.改进Canny算子与形态学融合的边缘检测[J].计算机工程与设计,2023,44(1):224-231. DOI:10.16208/j.issn1000-7024.2023.01.030 TANG S F, ZHAI S Q, TONG G M, et al. Improved edge detection by fusion of Canny operator and morphology[J]. Computer Engineering and Design, 2023, 44(1): 224-231. (in Chinese with English abstract) doi: 10.16208/j.issn1000-7024.2023.01.030
6	林宁宁,高心丹.DeepLab V3+改进的树木图像分割[J].计算机工程与设计,2023,44(1):232-239. DOI:10.16208/j.issn1000-7024.2023.01.031 LIN N N, GAO X D. Improved trees image segmentation based on DeepLab V3+[J]. Computer Engineering and Design, 2023, 44(1): 232-239. (in Chinese with English abstract) doi: 10.16208/j.issn1000-7024.2023.01.031
7	陈进,韩梦娜,练毅,等.基于U-Net模型的含杂水稻籽粒图像分割[J].农业工程学报,2020,36(10):174-180. DOI:10.11975/j.issn.1002-6819.2020.10.021 CHEN J, HAN M N, LIAN Y, et al. Segmentation of impurity rice grain images based on U-Net model[J]. Transactions of the CSAE, 2020, 36(10): 174-180. (in Chinese with English abstract) doi: 10.11975/j.issn.1002-6819.2020.10.021
8	胡春华,刘炫,计铭杰,等.基于SegNet与三维点云聚类的大田杨树苗叶片分割方法[J].农业机械学报,2022,53(6):259-264. DOI:10.6041/j.issn.1000-1298.2022.06.027 HU C H, LIU X, JI M J, et al. Single poplar leaf segmentation method based on SegNet and 3D point cloud clustering in field[J]. Transactions of the Chinese Society for Agricultural Machinery, 2022, 53(6): 259-264. (in Chinese with English abstract) doi: 10.6041/j.issn.1000-1298.2022.06.027
9	王俊强,李建胜,周华春,等.基于Deeplabv3+与CRF的遥感影像典型要素提取方法[J].计算机工程,2019,45(10):260-265. DOI:10.19678/j.issn.1000-3428.0053359 WANG J Q, LI J S, ZHOU H C, et al. Typical element extraction method of remote sensing image based on Deeplabv3+ and CRF[J]. Computer Engineering, 2019, 45(10): 260-265. (in Chinese with English abstract) doi: 10.19678/j.issn.1000-3428.0053359
10	慕涛阳,赵伟,胡晓宇,等.基于改进的DeepLabV3+模型结合无人机遥感的水稻倒伏识别方法[J].中国农业大学学报,2022,27(2):143-154. DOI:10.11841/j.issn.1007-4333.2022.02.14 MU T Y, ZHAO W, HU X Y, et al. Rice lodging recognition method based on UAV remote sensing combined with the improved DeepLabV3+ model[J]. Journal of China Agricultural University, 2022, 27(2): 143-154. (in Chinese with English abstract) doi: 10.11841/j.issn.1007-4333.2022.02.14
11	LI H L, TIAN J J, XIE Y X, et al. Performance evaluation of fusion techniques for cross-domain building rooftop segmenta-tion[C]//The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLIII- B3-2022. Nice, France: XXIV ISPRS Congress, 2022. DOI: 10.5194/isprs-archives-XLIII-B3-2022-501-2022 doi: 10.5194/isprs-archives-XLIII-B3-2022-501-2022
12	ZHANG X Y, SHI X W, ZHANG X B. Analysis of medical slide images processing using depth learning in histopath-ological studies of cerebellar cortex tissue[J]. International Journal of Advanced Computer Science and Applications, 2023, 14(1): 611-621. DOI: 10.14569/IJACSA.2023.0140167 doi: 10.14569/IJACSA.2023.0140167
13	黄渝萍,李伟生.医学图像融合方法综述[J].中国图象图形学报,2023,28(1):118-143. DOI:10.11834/jig.220603 HUANG Y P, LI W S. A review of medical image fusion methods[J]. Journal of Image and Graphics, 2023, 28(1): 118-143. (in Chinese with English abstract) doi: 10.11834/jig.220603
14	FU H X, MENG D, LI W H, et al. Bridgecrack semantic segmentation based on improved Deeplabv3+[J]. Journal of Marine Science and Engineering, 2021, 9(6): 671. DOI: 10.3390/jmse9060671 doi: 10.3390/jmse9060671
15	PHAM T. Semantic road segmentation using deep learning[C]//Proceedings of the 2020 Applying New Technology in Green Buildings (ATiGB). Da Nang, Vietnam: IEEE, 2021. DOI: 10.1109/ATiGB50996.2021.9423307 doi: 10.1109/ATiGB50996.2021.9423307
16	DIAKOGIANNIS F I, WALDNER F, CACCETTA P, et al. ResUNet-a: a deep learning framework for semantic segmentation of remotely sensed data[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2020, 162: 94-114. DOI: 10.1016/j.isprsjprs.2020.01.013 doi: 10.1016/j.isprsjprs.2020.01.013
17	CHEN L C, ZHU Y K, PAPANDREOU G, et al. Encoder-decoder with atrous separable convolution for semantic image segmentation[C]//European Conference on Computer Vision, Computer Vision-ECCV 2018. Cham, Germany: Springer, 2018. DOI: 10.1007/978-3-030-01234-2_49 doi: 10.1007/978-3-030-01234-2_49
18	CHEN L C, PAPANDREOU G, SCHROFF F, et al. Rethin-kingatrous convolution for semantic image segmentation[CP/OL]. arXiv, 2017, arXiv: 1706.05587.
19	王文成,蒋慧,乔倩,等.基于ResNet50网络的十种鱼类图像分类识别研究[J].农村经济与科技,2019,30(19):60-62. DOI:10.3969/j.issn.1007-7103.2019.19.028 WANG W C, JIANG H, QIAO Q, et al. Research on image classification and recognition of ten kinds of fish based on ResNet50 network[J]. Rural Economy and Science-Technology, 2019, 30(19): 60-62. (in Chinese) doi: 10.3969/j.issn.1007-7103.2019.19.028
20	孙晖,杨艾炯,李康博,等.基于深度学习的眼角膜图像自动化分析研究[J].吉林大学学报(信息科学版),2021,39(5):609-616. DOI:10.19292/j.cnki.jdxxp.2021.05.015 SUN H, YANG A J, LI K B, et al. Research on automated corneal image analysis based on deep learning[J]. Journal of Jilin University (Information Science Edition), 2021, 39(5): 609-616. (in Chinese with English abstract) doi: 10.19292/j.cnki.jdxxp.2021.05.015
21	余汛,王哲,景海涛,等.基于深度学习方法和RGB影像的玉米雄穗分割[J].浙江大学学报(农业与生命科学版),2021,47(4):451-463. DOI:10.3785/j.issn.1008-9209.2021.03.121 YU X, WANG Z, JING H T, et al. Maize tassel segmentation based on deep learning method and RGB image[J]. Journal of Zhejiang University (Agriculture & Life Sciences), 2021, 47(4): 451-463. (in Chinese with English abstract) doi: 10.3785/j.issn.1008-9209.2021.03.121

[1]	董成烨,李东方,冯槐区,龙思放,奚特,周芩安,王俊. 基于机器视觉和机器学习技术的浙贝母外观品质等级区分[J]. 浙江大学学报（农业与生命科学版）, 2023, 49(6): 881-892.
[2]	谢鹏尧,富昊伟,唐政,麻志宏,岑海燕. 基于RGB图像的冠层尺度水稻叶瘟病斑检测与抗性评估[J]. 浙江大学学报（农业与生命科学版）, 2021, 47(4): 415-428.
[3]	余汛,王哲,景海涛,金秀良,聂臣巍,白怡,王铮. 基于深度学习方法和RGB影像的玉米雄穗分割[J]. 浙江大学学报（农业与生命科学版）, 2021, 47(4): 451-463.
[4]	王永维，何焯亮，陈军，王俊，张羚玥，唐燕海. 碰撞气吹式杂交水稻授粉机结构与参数优化[J]. 浙江大学学报(农业与生命科学版), 2018, 44(1): 98-106.
[5]	胡泓　王光火. 金华长期定位试验中杂交水稻的钾素营养特征研究[J]. 浙江大学学报(农业与生命科学版), 2003, 29(3): 257-260.

Viewed

Full text

Abstract

Cited

Shared

Discussed