High-throughput identification of fusarium head blight resistance in wheat varieties using field robot-assisted imaging and deep learning techniques

文献类型： 外文期刊

第一作者： Liu, Tao

作者： Liu, Tao;Zhao, Yuanyuan;Sun, Yi;Yao, Zhaosheng;Sun, Chengming;Li, Tao;Liu, Tao;Zhao, Yuanyuan;Sun, Yi;Yao, Zhaosheng;Sun, Chengming;Li, Tao;Wang, Junchan;Chen, Chen;Zhong, Xiaochun;Liu, Shengping

作者机构：

关键词： Wheat; Fusarium head blight; Identification; Image processing; Deep learning

期刊名称：JOURNAL OF CLEANER PRODUCTION （ 影响因子：10.0； 五年影响因子：10.7 ）

ISSN： 0959-6526

年卷期： 2024 年 480 卷

页码：

收录情况： SCI

摘要： Fusarium head blight (FHB) has attracted much attention in food science and agriculture for its threat to wheat yields and food safety, due to the production of mycotoxins like deoxynivalenol (DON). Breeding of wheat varieties with improved FHB resistance is essential for controlling this disease. However, identifying resistance in different materials during variety selection remains time-consuming and labor-intensive. Therefore, this paper proposed a high-throughput method for the evaluation of FHB disease symptoms. It enabled the semiautomatic acquisition of images of individual wheat ears in a field environment with the aid of a field robot. The images obtained were semantically segmented to get a single wheat ear, from which the infected spikelets (ISs) were extracted, and then the disease degree was calculated. The results showed that the accuracy value of the individual wheat ear using DeepLabV3+ reached 0.996. The accuracy value of ISs was over 0.98. The mean Accuracy of this method for identifying the resistance of varieties was 0.967, and the precision of a single severity grade reached 0.980. The results indicate that the proposed method enables the acquisition and extraction of disease phenotype and rapid identification of resistance to FHB. The study also provides a reference for accurately identifying phenotypes of wheat ears in other field environments and for selecting and breeding other fungaltoxin-resistant varieties.

分类号：

  • 相关文献
作者其他论文 更多>>

微信小程序