结合改进胶囊网络与知识蒸馏的茶青分类方法研究

doi:10.13305/j.cnki.jts.2022.03.002

茶叶科学 ›› 2022, Vol. 42 ›› Issue (3): 387-396.doi: 10.13305/j.cnki.jts.2022.03.002

结合改进胶囊网络与知识蒸馏的茶青分类方法研究

陈星燃¹, 黄海松^1,*, 韩正功¹, 范青松¹, 朱云伟¹, 胡鹏飞^2,3

1.贵州大学现代制造技术教育部重点实验室,贵州贵阳 550025;
2.贵州装备制造职业学院,贵州贵阳 551400;
3.清镇红枫山韵茶场有限公司,贵州贵阳 551400

收稿日期:2021-11-30 修回日期:2022-02-16 出版日期:2022-06-15 发布日期:2022-06-17
通讯作者: *hshuang@gzu.edu.cn
作者简介:陈星燃,男,硕士研究生,主要从事深度学习与农业工程的交叉研究。
基金资助:
贵州省科技计划项目（黔科合支撑[2021]一般445、172、397,黔科合支撑[2022]一般165,黔科合基础[2020]1Y232）、贵州大学引进人才科研基金（贵大人基合字（2019）07号）、贵州省普通高等学校青年科技人才成长项目（黔教合KY 字[2021]096）

Research on the Classification Method of Tea Buds Combining Improved Capsule Network and Knowledge Distillation

CHEN Xingran¹, HUANG Haisong^1,*, HAN Zhenggong¹, FAN Qingsong¹, ZHU Yunwei¹, HU Pengfei^2,3

1. Key laboratory of Advanced Manufacturing Technology of the Ministry of Education, Guizhou University, Guiyang 550025, China;
2. Guizhou Vocational College of Equipment Manufacturing, Guiyang 551400, China;
3. Qingzhen Hongfeng Mountain Yun Tea Factory Co., Ltd, Guiyang 551400, China

Received:2021-11-30 Revised:2022-02-16 Online:2022-06-15 Published:2022-06-17

摘要/Abstract

摘要： 不同等级茶青的准确分类,对名优茶产业发展至关重要,采用传统感官审评方法进行分选会使结果存在一定的主观性。采集茶青图像建立数据集,结合幽灵注意力瓶颈层与胶囊网络提出一种新型网络模型：GA-CapsNet。通过基于线性衰减比例系数的成长知识蒸馏方法对该模型进行训练,在迁移教师模型参数矩阵的同时,使学生模型随着迭代自适应降低依赖程度。试验结果表明,对比其他同类算法,所提出的方法在小规模数据集上分类性能优异,精确率、召回率及F1-score分别为94.97%、95.51%、95.24%。本研究基于机器视觉与深度学习技术构建了一种GA-CapsNet模型,为解决茶青分类问题提供了一种新思路。

关键词: 胶囊网络, 知识蒸馏, 注意力模块, 茶青分类

Abstract: The accurate classification of different grades of tea buds is very important for the development of the famous tea industry. The use of traditional sensory evaluation methods for sorting makes the results subjective. In this research, a data set was established after tea leaf images were collected, and a new network model, GA-CapsNet, was proposed by combining the ghost attention bottleneck and capsule network. The model was trained by the method of growing knowledge distillation based on the linear decay scaling coefficient, while migrating the parameter matrix of teacher model, the student model was adaptively reduced with iteration. The experimental result shows that, compared with other similar algorithms, the proposed method had excellent classification performance on small-scale data sets. The accuracy, recall and F1-score were 94.97%, 95.51% and 95.24%, respectively. Here, a GA-CapsNet model based on machine vision and deep learning technology was established, which provided a new idea for solving the tea leaf classification problem.

Key words: capsule network, knowledge distillation, attention module, tea bud grading

中图分类号:

陈星燃, 黄海松, 韩正功, 范青松, 朱云伟, 胡鹏飞. 结合改进胶囊网络与知识蒸馏的茶青分类方法研究[J]. 茶叶科学, 2022, 42(3): 387-396. doi: 10.13305/j.cnki.jts.2022.03.002.

CHEN Xingran, HUANG Haisong, HAN Zhenggong, FAN Qingsong, ZHU Yunwei, HU Pengfei. Research on the Classification Method of Tea Buds Combining Improved Capsule Network and Knowledge Distillation[J]. Journal of Tea Science, 2022, 42(3): 387-396. doi: 10.13305/j.cnki.jts.2022.03.002.

参考文献

[1] 石亚丽, 朱荫, 马婉君, 等. 名优炒青绿茶挥发性成分研究进展[J]. 茶叶科学, 2021, 41(3): 285-301.
Shi Y L, Zhu Y, Ma W J, et al.Research progress on the volatile compounds of premium roasted green tea[J]. Journal of Tea Science, 2021, 41(3): 285-301.
[2] 刘元生, 刘方, 陈祖拥, 等. 贵州富锌硒茶产区地质环境条件与土壤元素地球化学特征[J]. 贵州大学学报(自然科学版), 2021, 38(5): 25-32.
Liu Y S, Liu F, Chen Z Y, et al.Geological environment conditions and geochemical characteristics of soil elements in zinc-selenium-rich tea producing areas in Guizhou[J]. Journal of Guizhou University (Natural Science), 2021, 38(5): 25-32.
[3] 陈青, 杨云, 王海燕. 贵州茶叶中微量锗含量测定[J]. 贵州大学学报(自然科学版), 2002, 19(2): 141-144.
Chen Q, Yang Y, Wang H Y.Determination of trace germanium content in Guizhou tea[J]. Journal of Guizhou University (Natural Science), 2002, 19(2): 141-144.
[4] 郭建军, 周艺, 王小英, 等. 贵州不同产区代表绿茶的品质特征及香气组分分析[J]. 食品工业科技, 2021, 42(5): 78-84.
Guo J J, Zhou Y, Wang X Y, et al.Analysis of quality features and aroma components in Guizhou representative green tea[J]. Food Industry Technology, 2021, 42(5): 78-84.
[5] 叶江华, 罗盛财, 张奇, 等. 武夷山不同茶园茶树茶青品质的差异[J]. 福建农林大学学报(自然科学版), 2017, 46(5): 495-501.
Ye J H, Luo S C, Zhang Q, et al.Difference of fresh tea leaf quality from different tea plantations in Wuyishan[J]. Journal of Fujian Agriculture and Forestry University (Natural Science Edition), 2017, 46(5): 495-501.
[6] 吴雪梅, 张富贵, 吕敬堂. 基于图像颜色信息的茶叶嫩叶识别方法研究[J]. 茶叶科学, 2013, 33(6): 584-589.
Wu X M, Zhang F G, Lv J T.Research on recognition of tea tender leaf based on image color information[J]. Journal of Tea Science, 2013, 33(6): 584-589
[7] 张怡, 赵珠蒙, 王校常, 等. 基于ResNet卷积神经网络的绿茶种类识别模型构建[J]. 茶叶科学, 2021, 41(2): 261-271.
Zhang Y, Zhao Z M, Wang X C, et al.Construction of green tea recognition model based on ResNet convolutional neural network[J]. Journal of Tea Science, 2021, 41(2): 261-271.
[8] 汪建, 杜世平. 基于颜色和形状的茶叶计算机识别研究[J]. 茶叶科学, 2008, 28(6): 420-424.
Wang J, Du S P.Identification investigation of tea based on HSI color space and figure[J]. Journal of Tea Science, 2008, 28(6): 420-424.
[9] 毛腾跃, 黄印, 文晓国, 等. 基于多特征与多分类器的鲜茶叶分类研究[J]. 中国农机化学报, 2020, 41(12): 75-83.
Mao T Y, Huang Y, Wen X G, et al.Research on classification of fresh tea based on multiple features and multiple classifiers[J]. Journal of Chinese Agricultural Mechanization, 2020, 41(12): 75-83.
[10] 张晴晴. 基于卷积神经网络的茶树嫩芽识别研究[D]. 合肥: 安徽农业大学, 2020.
Zhang Q Q.Research on the identification of tea tree sprouts based on convolutional neural network [D]. Hefei: Anhui Agricultural University, 2020.
[11] 张蕴. 茶叶嫩芽图像识别方法研究[D]. 合肥: 安徽农业大学, 2020.
Zhang Y.Research on image recognition method of tea sprouts [D]. Hefei: Anhui Agricultural University, 2020.
[12] Xu M, Wang J, Gu S.Rapid identification of tea quality by E-nose and computer vision combining with a synergetic data fusion strategy[J]. Journal of Food Engineering, 2019, 241: 10-17.
[13] Sabour S, Frosst N, Hinton G E.Dynamic routing between capsules[C]//Luxburg U V, Guyon I, Bengio, et al. Proceedings of the 31st International Conference on Neural Information Processing Systems. New York: Curran Associates Inc., 2017: 3859-3869.
[14] Hinton G E, Sabour S, Frosst N. Matrix capsules with em routing [C/OL]//ICLR Organizing Committee. ICLR2018 Conference Blind Submission,2018[2022-11-26].
[15] Wang Q, Wu B, Zhu P, et al.ECA-Net: efficient channel attention for deep convolutional neural networks[J]. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020: 11531-11539.
[16] 魏铖磊, 南新元, 李成荣, 等. 一种具有多尺度感受视野注意力机制的生活垃圾单阶段目标检测方法[J]. 环境工程, 2022, 40(1): 175-183.
Wei C L, Nan X Y, Li C R, et al.A single-stage object detection method for domestic garbage based on multi-scale receptive filed attention mechanism[J]. Environmental Engineering: 2022, 40(1): 175-183.
[17] 徐岩, 李晓振, 吴作宏, 等. 基于残差注意力网络的马铃薯叶部病害识别[J]. 山东科技大学学报(自然科学版), 2021, 40(2): 76-83.
Xu Y, Li X Z, Wu Z H, et al.Potato leaf disease recognitition via residual atention network[J]. Journal of Shandong University of Science and Technology (Natural Science), 2021, 40(2): 76-83.
[18] Zhang S, Zhang S, Zhang C, et al.Cucumber leaf disease identification with global pooling dilated convolutional neural network[J]. Computers and Electronics in Agriculture, 2019, 162: 442-430.
[19] Toraman S, Alakus T B, Turkoglu I.Convolutional capsnet: a novel artificial neural network approach to detect COVID-19 disease from X-ray images using capsule networks[J]. Chaos, Solitons & Fractals, 2020, 140: 110122. doi: 10.1016/j.chaos.2020.110122.
[20] Misra D.Mish: a self regularized non-monotonic activation function[J]. BMVC 2020, 2020. doi: 10.48550/arXiv.1908.
08681.
[21] Hu G, Yang X, Zhang Y, et al.Identification of tea leaf diseases by using an improved deep convolutional neural network[J]. Sustainable Computing: Informatics and Systems, 2019, 24: 100353. doi: 10.1016/j.suscom.2019.100353.

结合改进胶囊网络与知识蒸馏的茶青分类方法研究

Research on the Classification Method of Tea Buds Combining Improved Capsule Network and Knowledge Distillation

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

Metrics

本文评价

推荐阅读 10

[1]	王留彬, 黄丽蕴, 滕翠琴, 吴立赟, 成浩, 于翠平, 王丽鸳. 梧州茶树种质资源的遗传多样性及亲缘关系分析[J]. 茶叶科学, 2022, 42(5): 601-609.
[2]	武警, 陈楠楠, 韩梦琳, 陈高, 厉伟伟, 张蜀香, 蒋晓岚. 茶树根系耐铝促生内生细菌的分离鉴定及其特性研究[J]. 茶叶科学, 2022, 42(5): 610-622.
[3]	高健健, 陈丹, 彭佳堃, 吴文亮, 蔡良绥, 蔡亚威, 田军, 万云龙, 孙威江, 黄艳, 王哲, 林智, 戴伟东. 基于代谢组学的云南白茶与福鼎白茶化学成分比较分析[J]. 茶叶科学, 2022, 42(5): 623-637.
[4]	周汉琛, 杨霁虹, 徐玉婕, 吴琼, 雷攀登. 香叶醇生物合成相关基因NUDX1的进化分析[J]. 茶叶科学, 2022, 42(5): 638-648.
[5]	陈琪予, 马建强, 陈杰丹, 陈亮. 利用图像特征分析茶树成熟叶表型的遗传多样性[J]. 茶叶科学, 2022, 42(5): 649-660.
[6]	李艳春, 汪航, 李兆伟, 叶菁, 王义祥. 几种改良措施对酸化茶园土壤理化性质和微生物群落结构的影响[J]. 茶叶科学, 2022, 42(5): 661-671.
[7]	王峰, 陈玉真, 吴志丹, 尤志明, 余文权, 俞晓敏, 杨贞标. 有机管理模式对茶园土壤真菌群落结构及功能的影响[J]. 茶叶科学, 2022, 42(5): 672-688.
[8]	孙悦, 吴俊, 韦朝领, 刘梦月, 高晨曦, 张灵枝, 曹士先, 余顺甜, 金珊, 孙威江. 抗小贯松村叶蝉和茶棍蓟马的茶树种质筛选及其抗性相关因素分析[J]. 茶叶科学, 2022, 42(5): 689-704.
[9]	陈宇宏, 高颖, 韩震, 尹军峰. 不同种质茶叶籽皂素含量及组成分析[J]. 茶叶科学, 2022, 42(5): 705-716.
[10]	陈慧, 杨丽玲, 陈金华, 黄建安, 龚雨顺, 李适. 控温渥堆对黑毛茶香气品质的影响[J]. 茶叶科学, 2022, 42(5): 717-730.
[11]	陈珂, 王苑竹, 杨晓颖, 张冬英, 朱强强. 壳聚糖复合β-乳球蛋白负载EGCG纳米粒的制备及其对糖尿病小鼠血糖的影响[J]. 茶叶科学, 2022, 42(5): 731-739.
[12]	李峥, 刘锭, 霍增辉, 陈富桥. 中国与RCEP成员国茶叶贸易竞争性与互补性分析[J]. 茶叶科学, 2022, 42(5): 740-752.
[13]	俞蓉欣, 郑芹芹, 陈红平, 张劲松, 张相春. 儿茶素生物医用纳米材料研究进展[J]. 茶叶科学, 2022, 42(4): 447-462.
[14]	王玉源, 刘任坚, 刘少群, 舒灿伟, 孙彬妹, 郑鹏. 茶树R2R3-MYB转录因子CsTT2表达分析及功能初步鉴定[J]. 茶叶科学, 2022, 42(4): 463-476.
[15]	李晶, 林彩容, 黄艳, 邓旭铭, 王艺清, 孙威江. 茶多酚对农杆菌介导的植物遗传转化体系的影响[J]. 茶叶科学, 2022, 42(4): 477-490.