Wheat seed classification based on improved lightweight EfficientNet-V2 model

doi:10.13733/j.jcam.issn.2095-5553.2024.09.017

Abstract

Abstract:

Wheat seeds are often mixed with other seeds such as oats and barley，and how to classify seeds of sufficient purity is an important problem. In order to address this issue， a seed classification method based on an improved EfficientNet-V2 model is proposed，and the enhanced network is named CA-EfficientNet-V2_xs. Firstly，a dataset is constructed by purchasing commonly used wheat seeds（including common impurities such as oats and barley）. Secondly， in order to expedite training and overcome the issue of insufficient data in the self.made dataset，transfer learning is adopted. Thirdly，in order to assist the model in more accurately locating and identifying the target of interest， the Coordinate Attention（CA） mechanism is adopted to replace the SE attention mechanism. Finally，the network structure is streamlined to reduce the model size and enhance training speed. The experimental results show that the classification accuracy of the improved model reaches 99. 7%，which is 1. 3% higher than that of the network before the improvement. Compared with 78 MB in the EfficientNet-V2_s model，the improved model size is reduced to 3. 8 MB and the model size is reduced. The improved model is faster than mainstream networks.

Key words: wheat seed, deep learning, attention mechanism, transfer learning, EfficientNet-V2 model

摘要：

小麦种子常混有燕麦、大麦等其他种子，如何分类足够纯度的种子是一个重要问题。为解决种子纯度问题，提出一种基于改进的 EfficientNet-V2模型的种子分类方法，将其命名为 CA-EfficientNet-V2_xs。首先，通过购买常用小麦种子（含有常见的燕麦、大麦等杂种），自制数据集；其次，为加快训练以及针对自制数据集数量不足的问题，采用迁移学习的方法；再次，为更好地帮助模型更加精准地定位和识别感兴趣的目标，将采用 Coordinate Attention（CA）注意力机制来替换 SE注意力机制；最后，通过精简网络结构使模型更小、训练速度更快。试验表明，改进后模型的分类准确率达到 99. 7%，比未改进之前的网络分类准确率提升 1. 3%；与 EfficientNet-V2_s模型的 78 MB相比，改进后模型大小降至 3. 8 MB，模型大小降低；改进后的模型速度比主流网络更快。

关键词: 小麦种子, 深度学习, 注意力机制, 迁移学习, EfficientNet-V2模型

CLC Number:

Han Pengfei, Song Qijiang, Jia Mengshi. Wheat seed classification based on improved lightweight EfficientNet-V2 model[J]. Journal of Chinese Agricultural Mechanization, 2024, 45(9): 111-117.

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.

References

［ 1］姜楠，韩一军，李雪 .中国小麦种业发展研究［ J］.中国种业， 2013（10）：1-4.

［ 2］任庭玺 .海晏县农田野燕麦影响小麦产量的调查［ J］.青海农牧业， 2010（2）：8，16.

［ 3］Eryigit R，Tugrul B. Performance of various deep.learning networks in the seed classification problem［J］. Symmetry，2021，13（10）：1892.

［ 4］Zhang J，Dai L，Cheng F. Identification of corn seeds with different freezing damage degree based on hyperspectral reflectance imaging and deep learning method［J］. Food Analytical Methods，2021，14：389-400.

［ 5］Kiratiratanapruk K，Temniranrat P，Sinthupinyo W，et al. Development of paddy rice seed classification process using machine learning techniques for automatic grading machine［J］. Journal of Sensors，2020（1）：7041310.

［ 6］Gulzar Y，Hamid Y，Soomro A B，et al. A convolution neural network.based seed classification system［J］. Symmetry，2020，12（12）：2018.

［ 7］袁国武，刘建成，刘鸿瑜，等 .基于 ResNeSt的烟支外观缺陷分类［ J］.云南大学学报（自然科学版）， 2022，44（3）：464-470.

Yuan Guowu， Liu Jiancheng， Liu Hongyu， et al. Classification of cigarette appearance defects based on ResNeSt［J］. Journal of Yunnan University（Natural Sciences Edition），2022，44（3）：464-470.

［ 8］张涛 .基于随机森林和支持向量机在小麦种子分类中的比较研究［ J］.河西学院学报， 2020，36（2）：33-40.

Zhang Tao. A comparative study on the application of random forest and support vector machine to wheat seed classification［J］. Journal of Hexi University，2020，36（2）：33-40.

［ 9］张宇卓，王德成，方宪法，等 .基于 CNN的玉米种子内部裂纹图像检测系统［ J］.农业机械学报， 2022，53（5）： 309-315.

Zhang Yuzhuo， Wang Decheng， Fang Xianfa， et al. Image detection system of corn seed internal crack based on CNN［J］. Transactions of the Chinese Society for Agricultural Machinery，2022，53（5）：309-315.

［10］樊湘鹏，周建平，许燕，等 .基于改进卷积神经网络的复杂背景下玉米病害识别［ J］.农业机械学报， 2021，52（3）：210-217.

Fan Xiangpeng， Zhou Jianping， Xu Yan， et al. Corn disease recognition under complicated background based on improved convolutional neural network［J］. Transactions of the Chinese Society for Agricultural Machinery，2021， 52（3）：210-217.

［11］胡定一 .基于深度学习的缺陷柑橘分类识别检测方法研究［ D］.武汉：华中农业大学， 2021.

［12］胡志伟，杨华，黄济民，等 .基于注意力残差机制的细粒度番茄病害识别［J］.华南农业大学学报， 2019，40（6）：124-132.

Hu Zhiwei，Yang Hua，Huang Jimin，et al. Fine.grained tomato disease recognition based on attention residual mechanism［J］. Journal of South China Agricultural University，2019，40（6）：124-132.

［13］Velesaca H O，Mira R，Suárez P L，et al. Deep learning based corn kernel classification［C］. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops，2020：66-67.

［14］Krizhevsky A， Sutskever I， Hinton G E. Imagenet classification with deep convolutional neural networks［J］. Advances in Neural Information Processing Systems， 2012，25.

［15］廉小亲，成开元，安飒，等 .基于深度学习和迁移学习的水果图像分类［ J］.测控技术， 2019，38（6）：15-18.

Lian Xiaoqin，Cheng Kaiyuan，An Sa，et al. Fruit image classification based on deep learning and transfer learning［J］. Measurement & Control Technology，2019， 38（6）：15-18.

［16］Javanmardi S， Ashtiani S H M， Verbeek F J， et al. Computer.vision classification of corn seed varieties using deep convolutional neural network［J］. Journal of Stored Products Research，2021，92：101800.

［17］Gulzar Y，Hamid Y，Soomro A B，et al. A convolution neural network.based seed classification system［J］. Symmetry， 2020，12（12）：2018.

[1]	Zhao Shuang, Yu Yongqiang, Miao Yubin, Liu Kexin. Research of impurity detection of green vegetable based on improved Mask R-CNN [J]. Journal of Chinese Agricultural Mechanization, 2024, 45(9): 77-82.
[2]	Yang Haoyong, Wang Chaozhu, Liu Haoyi, Guan Xintong, Liu Yingying, Ding Yongqian. Research on a rapid detection method for seed breakage rate [J]. Journal of Chinese Agricultural Mechanization, 2024, 45(9): 104-110.
[3]	Li Zhichen, Ling Xiujun, Li Hongqiu. Engine misfire fault diagnosis based on sound deep learning [J]. Journal of Chinese Agricultural Mechanization, 2024, 45(9): 134-140.
[4]	Sang Yinan, Xu Zenglai, Wang Qiong, Ge Haitao, Wang Dianguang. Design of a wheeled flower picking robot: Taking Abelmoschus manihot (Linn.) Medicus as an example [J]. Journal of Chinese Agricultural Mechanization, 2024, 45(9): 202-208.
[5]	Zhang Chuandong, Qi Lu, Ding Huali. Detection method of multi variety grape cluster based on improved YOLOv8n deep learning algorithm [J]. Journal of Chinese Agricultural Mechanization, 2024, 45(9): 220-226.
[6]	Li Bangguo, Wang Hui, Song Yang, Ren Zhiwei, Liu Yuehua, Xu Lecheng. Stereo visual perception system based on unmanned wheat harvester [J]. Journal of Chinese Agricultural Mechanization, 2024, 45(9): 244-249.
[7]	Song Xiaolu, Feng Juan, Liang Xiangyu, Zhou Xixing. Recognition and counting algorithm of underwater sea cucumbers based on YOLOv4 network [J]. Journal of Chinese Agricultural Mechanization, 2024, 45(9): 258-264.
[8]	Li Jie, , Gao Shangbing, Yu Jiyuan, , Chen Xin, Li Shicong, Yuan Xingxing. Identification method of resistance to mung bean leaf spot disease based on PMMS-Net and chlorophyll fluorescence imaging [J]. Journal of Chinese Agricultural Mechanization, 2024, 45(8): 210-216.
[9]	Xu Zhenhui, Li Xiaojuan. Research on tree trunk detection and navigation line fitting algorithm in orchard [J]. Journal of Chinese Agricultural Mechanization, 2024, 45(8): 217-222.
[10]	LiuXing, Gu Jinan, Huang Zedong, Zhang Wenhao, Zhang Wei. Research on apple point cloud semantic segmentation based on deep learning [J]. Journal of Chinese Agricultural Mechanization, 2024, 45(8): 223-227.
[11]	Su Fei, , Wang Guanghui, Shi Yanxia, Jia Ran, Yan Yinfa, , Zu Linlu, . Leaf disease segmentation model of greenhouse tomatoes based on ResUnet with attention mechanism [J]. Journal of Chinese Agricultural Mechanization, 2024, 45(8): 228-233.
[12]	Hu Guoyu, , Liu Guang, Zhou Xingguang, Dong Yalan, Zhou Jianping, . Detection method of grape disease in orchard environment based on Swin-TDL algorithm [J]. Journal of Chinese Agricultural Mechanization, 2024, 45(8): 234-239.
[13]	Feng Xiao, , Zhang Hui, , Liu Yunchao, Zhang Wei, , Li Xiaohong, , Ma Zhongjie, . Detection of grape cluster in stone hardening stage based on improved YOLOv5s [J]. Journal of Chinese Agricultural Mechanization, 2024, 45(8): 240-245.
[14]	Zhou Sijie, Liu Tianqi, Chen Tianhua. Research on rice disease recognition based on improved YOLOv5 algorithm [J]. Journal of Chinese Agricultural Mechanization, 2024, 45(8): 246-253.
[15]	Wang Ketao, Chen Shifeng, Chen Gui, Wei Jin, Meng Liwen, Chen Quancheng. Detection method of garden spherical hedge based on YOLOv5s [J]. Journal of Chinese Agricultural Mechanization, 2024, 45(8): 262-269.