基于注意力机制和迁移学习的小样本茶叶病害识别

doi:10.13733/j.jcam.issn.2095-5553.2024.10.038

摘要/Abstract

摘要： 为提高茶叶病害识别的准确性，提出一种基于注意力机制和迁移学习的小样本茶叶病害图像识别方法。首先，通过对训练集图像随机旋转、随机翻转、随机色彩以及随机亮度调整操作，扩充训练集样本，旨在降低网络模型的过拟合风险。其次，引入卷积块注意力机制(CBAM)对ResNet50网络模型进行优化，使其能够更加精准地聚焦于茶叶病害的特征。最后，借助植物病害数据集对ResNet50模型进行预训练，并将预训练后的模型参数迁移到茶叶病害数据集进行训练。试验结果表明：扩充后的数据集识别准确率较原数据集提高7.85%，进行二次迁移学习后识别准确率又提高3.26%，再添加CBAM注意力机制后识别准确率又提高0.81%。在对8种茶树病害进行验证时，所提方法的样本识别率与原始模型相比由77.24%提高至89.16%。改进后的模型具有更好的特征提取能力，能够有效提高小样本茶叶病害的识别准确率。

关键词: 图像识别, 深度学习, 注意力机制, 迁移学习, 茶叶病害

Abstract: In order to improve the accuracy of tea disease recognition, a recognition method of small sample tea disease image based on attention mechanism and transfer learning was proposed. Firstly, the training set samples were expanded by random rotation, random flip, random color and random brightness adjustment operations to reduce the overfitting risk of the network model. Secondly, Convolutional Block Attention Module (CBAM) was introduced to optimize the ResNet50 network model so that it could focus more accurately on the characteristics of tea diseases. Finally, the ResNet50 model was pre‑trained with the help of plant disease data set, and the pre‑trained model parameters were transferred to the tea disease data set for training. The experimental results show that the recognition accuracy of the expanded data set is 7.85% higher than that of the original data set, the recognition accuracy is 3.26% higher after two transfer learning, and the recognition accuracy is 0.81% higher after adding CBAM attention mechanism. Compared with the original model, the sample recognition rate of the proposed method is increased from 77.24% to 89.16% when 8 tea tree diseases are verified. The improved model has better feature extraction ability, which can effectively improve the recognition accuracy of tea disease in small samples.

Key words: image recognition, deep learning, attention mechanism, transfer learning, tea disease

中图分类号:

张莉, 杨明辉, 孙嘉成. 基于注意力机制和迁移学习的小样本茶叶病害识别[J]. 中国农机化学报, 2024, 45(10): 262-268.

Zhang Li, Yang Minghui, Sun Jiacheng. Identification method of small sample tea leaf diseases based on attention mechanism and transfer learning[J]. Journal of Chinese Agricultural Mechanization, 2024, 45(10): 262-268.

参考文献

[ 1 ] Xue Z Y, Xu R J, Bai D, et al. YOLO-Tea: A tea disease detection model improved by YOLOv5 [J]. Forests, 2023(14): 415.
[ 2 ] Hu G S, Yang X W, 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: 100353.1-100353.8.
[ 3 ] Hu G S, Wu H Y, Zhang Y, et al. A low shot learning method for tea leaf's disease identification [J]. Computers and Electronics in Agriculture, 2019, 163: 104852.
[ 4 ] Yuan X F, Li L, Wang Y L. Nonlinear dynamic soft sensor modeling with supervised long short‑term memory network [J]. IEEE Transactions on Industrial Informatics, 2020, 16(5): 3168-3176.
[ 5 ] Qiao J F, Hu Z Q, Li W J. Soft measurement modeling based on chaos theory for biochemical oxygen demand (BOD) [J]. Water, 2016, 8(581): 1-21.
[ 6 ] 翁杨, 曾睿, 吴陈铭, 等. 基于深度学习的农业植物表型研究综述[J]. 中国科学, 2019, 49(6): 698-716.
[ 7 ] Gavhale K R， Gawande U. An overview of the research on plant leaves disease detection using image processing techniques [J]. IOSR Journal of Computer Engineering, 2014, 16(1): 10-16.
[ 8 ] 翟肇裕, 曹益飞, 徐焕良, 等. 农作物病虫害识别关键技术研究综述[J].农业机械学报, 2021, 52(7): 1-18.
Zhai Zhaoyu, Cao Yifei, Xu Huanliang, et al. Review of key techniques for crop disease and pest detection [J]. Transactions of the Chinese Society of Agricultural Machinery, 2021, 52(7): 1-18.
[ 9 ] Chaudhary A, Kolhe S, Kamal R. An improved random forest classifier for multi‑class classification [J]. Information Processing in Agriculture, 2016, 3(4): 215-222.
[10] Tetila E C, Machado B B, Belete N, et al. Identification of soybean foliar diseases using unmanned aerial vehicle images [J]. IEEE Geoscience and Remote Sensing Letters, 2017: 1-5.
[11] Kiani E, Mamedov T. Identification of plant disease infection using soft‑computing: Application to modern botany [J]. Procedia Computer Science, 2017: 893-900.
[12] 何雨霜, 王琢, 王湘平, 等. 深度学习在农作物病害图像识别中的研究进展[J].中国农机化学报, 2023, 44(2): 148-155.
He Yushuang, Wang Zhuo, Wang Xiangping, et al. Research progress of deep learning in crop disease image recognition [J]. Journal of Chinese Agricultural Mechanization, 2023, 44(2): 148-155.
[13] Zhang X H, Qiao Y, Meng F F, et al. Identification of maize leaf diseases using improved deep convolutional neural networks [J]. IEEE Access, 2018: 2844405.
[14] Ferentinos K P. Deep learning models for plant disease detection and diagnosis [J]. Computers and Electronics in Agriculture. 2018, 145: 311-318.
[15] Lu Y, Yi S J, Zeng N Y, et al. Identification of rice diseases using deep convolutional neural networks [J]. Neurocomputing, 2017, 267: 378-384.
[16] 李博, 江朝晖, 洪石兰, 等. 基于边缘智能的茶叶病害识别[J]. 中国农机化学报, 2022, 43(6): 175-180.
Li Bo, Jiang Zhaohui, Hong Shilan, et al. Tea leaf diseases recognition based on edge intelligence [J]. Journal of Chinese Agricultural Mechanization, 2022, 43(6): 175-180.
[17] Pan S J, Yang Q. A survey on transfer learning [J]. IEEE Transactions on Knowledge and Data Engineering, 2010, 22(10): 1345-1359.
[18] Weiss K, Khoshgoftaar T M, Wang D D. A survey of transfer learning [J]. Journal of Big Data, 2016, 3(9): 1-40.
[19] Sladojevic S, Arsenovic M, Anderla A, et al. Deep neural networks based recognition of plant diseases by leaf image classification [J]. Computational Intelligence and Neuroscience, 2016: 3289801.
[20] Mohanty S P, Hughes D P, Salathé M. Using deep learning for image‑based plant disease detection [J]. Frontiers in Plant Science, 2016: 01419.
[21] 龙满生, 欧阳春娟, 刘欢, 等. 基于卷积神经网络与迁移学习的油茶病害图像识别[J]. 农业工程学报, 2018, 34(18): 194-201.
Long Mansheng, Ouyang Chunjuan, Liu Huan, et al. Image recognition of Camellia oleifera diseases based on convolutional neural network & transfer learning [J]. Transactions of the Chinese Society of Agricultural Engineering, 2018, 34(18): 194-201.
[22] 钱嘉鑫, 余鹏飞, 李海燕, 等. 基于特征融合与注意力机制的野生菌细粒度分类[J]. 激光与光电子学进展, 2023, 60(4): 110-119.
Qian Jiaxin, Yu Pengfei, Li Haiyan, et al. Fine‑grained classification of wild mushrooms based on feature fusion and attention mechanism [J]. Laser & Optoelectronics Progress, 2023, 60(4): 110-119.
[23] 孙俊，朱伟栋，罗元秋，等．基于改进MobileNet-V2的田间农作物叶片病害识别[J]. 农业工程学报, 2021, 37(22): 161-169.
Sun Jun, Zhu Weidong, Luo Yuanqiu, et al. Recognizing the diseases of crop leaves in fields using improved Mobilenet-V2 [J]. Transactions of the Chinese Society of Agricultural Engineering, 2021, 37(22): 161-169.
[24] 孙云云, 江朝晖, 董伟, 等. 基于卷积神经网络和小样本的茶树病害图像识别[J]. 江苏农业学报, 2019, 35(1): 48-55.
[25] He K M, Zhang X Y, Ren S Q, et al. Deep residual learning for image recognition [C]. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016: 770-778.

[1]	赵爽, 俞永强, 苗玉彬, 刘可心 . 基于改进 Mask R-CNN的青菜杂质检测研究[J]. 中国农机化学报, 2024, 45(9): 77-82.
[2]	韩鹏飞, 宋其江, 贾梦实. 基于改进轻量化 EfficientNet-V2模型的小麦种子分类[J]. 中国农机化学报, 2024, 45(9): 111-117.
[3]	李志臣, 凌秀军, 李鸿秋. 基于声音深度学习的发动机失火故障诊断[J]. 中国农机化学报, 2024, 45(9): 134-140.
[4]	桑一男 , 徐增莱 , 汪琼 , 葛海涛 , 王殿广 . 一种轮式采花机器人设计——以黄蜀葵为例[J]. 中国农机化学报, 2024, 45(9): 202-208.
[5]	张传栋, 亓璐, 丁华立. 基于改进 YOLOv8n模型的多品种葡萄簇检测方法[J]. 中国农机化学报, 2024, 45(9): 220-226.
[6]	李邦国, 王辉, 宋杨, 任志伟, 刘跃华, 徐乐程. 基于无人驾驶小麦收割机立体视觉感知系统[J]. 中国农机化学报, 2024, 45(9): 244-249.
[7]	宋小鹿 , 冯娟 , 梁翔宇 , 周玺兴 . 基于 YOLOv4的水下海参检测与计数算法[J]. 中国农机化学报, 2024, 45(9): 258-264.
[8]	李洁, 高尚兵, 余骥远, 陈新, 李士丛, 袁星星. 基于PMMS-Net和叶绿素荧光成像的绿豆叶斑病抗性鉴定方法[J]. 中国农机化学报, 2024, 45(8): 210-216.
[9]	许贞辉, 李晓娟. 果园树干检测与导航线拟合算法研究[J]. 中国农机化学报, 2024, 45(8): 217-222.
[10]	刘星, 顾寄南, 黄则栋, 张文浩, 张伟. 基于深度学习的苹果点云语义分割方法研究[J]. 中国农机化学报, 2024, 45(8): 223-227.
[11]	苏斐, 王光辉, 史艳霞, 贾然, 闫银发, 祖林禄, . 融合通道注意力机制的ResUnet作物病斑分割模型[J]. 中国农机化学报, 2024, 45(8): 228-233.
[12]	胡国玉, 刘广, 周星光, 董娅兰, 周建平, . 基于Swin-TDL算法的果园环境下葡萄病害检测方法[J]. 中国农机化学报, 2024, 45(8): 234-239.
[13]	冯晓, 张辉, 刘运超, 张微, 李小红, 马中杰, . 基于改进YOLOv5s的硬核期葡萄簇检测[J]. 中国农机化学报, 2024, 45(8): 240-245.
[14]	周思捷, 刘天奇, 陈天华. 基于改进YOLOv5算法的水稻病害识别研究[J]. 中国农机化学报, 2024, 45(8): 246-253.
[15]	王克涛, 陈世锋, 陈贵, 韦锦, 蒙丽雯, 陈泉成. 基于YOLOv5s的园林球形绿篱检测方法[J]. 中国农机化学报, 2024, 45(8): 262-269.