欢迎访问《中国农学通报》，
中国农学通报

中国农学通报 ›› 2025, Vol. 41 ›› Issue (2): 109-116.doi: 10.11924/j.issn.1000-6850.casb2024-0387

• 植物保护 • 上一篇    下一篇

基于迁移学习ResNet-18的水稻病虫害识别研究

张志从1(), 崔东1,2, 郭金锋1, 吾木提·艾山江1,2(), 李亮3   

  1. 1 伊犁师范大学资源与环境学院，新疆伊宁 835000
    2 伊犁师范大学资源与生态研究所，新疆伊宁 835000
    3 河北工业大学人工智能与数字科学学院，天津 300401
  • 收稿日期:2024-06-17 修回日期:2024-09-27 出版日期:2025-01-13 发布日期:2025-01-13
  • 通讯作者:
    吾木提·艾山江，男，1992年出生，新疆伊宁人，讲师，硕士，主要从事农业遥感研究。通信地址：835000 新疆伊犁州伊宁市解放路伊犁师范大学，E-mail：。
  • 作者简介:

    张志从，男，2000年出生，河北承德人，研究方向：高光谱数据建模与估算。通信地址：835000 新疆伊犁州伊宁市解放路伊犁师范大学，E-mail：。

  • 基金资助:
    伊犁师范大学科研项目“伊犁河谷稻田土壤微生物群落地理分布及其驱动因子研究”(2022YSYY003); 伊犁哈萨克自治州科技计划项目(YJC2024A05)

Research on Identification of Rice Disease and Pest Based on Transfer Learning and ResNet-18

ZHANG Zhicong1(), CUI Dong1,2, GUO Jinfeng1, UMUT Hasan1,2(), LI Liang3   

  1. 1 College of Resources and Environment, Yili Normal University, Yining, Xijiang 835000
    2 Institute of Resources and Ecology, Yili Normal University, Yining, Xijiang 835000
    3 School of Artificial Intelligence, Hebei University of Technology, Tianjin 300401
  • Received:2024-06-17 Revised:2024-09-27 Published:2025-01-13 Online:2025-01-13

RichHTML

3

PDF (PC)

19

摘要：

为提高水稻病虫害图像自动识别的准确性并进而有效指导农业病虫害防治工作。本研究采用迁移学习和ResNet-18模型相结合的方法，通过整理网络开源植物病害数据，获取水稻白叶枯病、稻瘟病和东格鲁病等9种水稻病虫害和1种健康叶片的图像数据，共计11414张用于模型训练，同时拆分30%的数据作为测试集。并在ResNet-18、GoogLeNet、VGG-16、MobileNet-v2等6种预训练模型的基础上，对迁移模型进行一系列的参数调整，以优化迁移学习模型。结果表明：（1）在模型训练参数一致情况下，本研究模型ResNet-18较MobileNet-v2、AxeNet、VGG-16、、GoogLeNet、SqueezeNet和原ResNet-18模型相比，不仅验证精度显著提高，损失值还最低，模型最终精度为96.97%；（2）相比于原模型，所有迁移学习后模型的训练精度提高较为明显，所提高精度分布在5.03%~13.90%。优化后的训练模型具有识别速度快、准确率提高的特点，可以准确、快速地识别出病害类型，为作物病害的自动诊断提供支撑。

关键词: 水稻, 深度学习, 病虫害, 迁移学习, ResNet-18, 图像识别

Abstract:

The study aims to improve the automatic recognition of rice pest and disease images and better guide agricultural pest and disease control. Using a combination of transfer learning and ResNet-18 model, we organized open source plant disease data on the internet, and obtained images of 9 rice pests and diseases, including bacterial blight, blast and Tungro, as well as a healthy leaf as the research objects. 11414 cleaned images were selected to establish a dataset for model training, and the 30% dataset was split as the test set. On the basis of six pre trained models such as ResNet-18, GoogLeNet, VGG-16, and MobileNet-v2, a series of parameter adjustments were made to the transfer model. The results show that: (1) under the consistent training parameters, the proposed model ResNet-18 has significantly higher validation accuracy and lowest loss value compared with MobileNet-v2, AxeNet, VGG-16, GoogLeNet, SqueezeNet, and the original ResNet-18 model. The final accuracy of the model is 96.97%. (2) Compared with the original model, the training accuracy of all transferred learning models has been improved significantly, with the improved accuracy ranging from 5.03% to 13.90%. The optimized training model has the characteristics of fast recognition speed and improved accuracy, which can accurately and quickly identify the type of crop disease, providing support for the automatic diagnosis of crop diseases.

Key words: rice, deep learning, diseases and pests, transfer learning, ResNet-18, image recognition