Early Recognition of Sclerotinia Stem Rot on Oilseed Rape by Hyperspectral Imaging Combined With Deep Learning
LIANG Wan-jie1, FENG Hui2, JIANG Dong3, ZHANG Wen-yu1, 4, CAO Jing1, CAO Hong-xin1*
1. Institute of Agricultural Information, Jiangsu Academy of Agricultural Sciences, Nanjing 210014, China
2. Institute of Plant Protection, Jiangsu Academy of Agricultural Sciences, Nanjing 210014, China
3. Institute of Geographical Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, China
4. School of Agricultural Engineering, Jiangsu University, Zhenjiang 212013, China
Abstract：The sclerotinia stem rot on oilseed rapeis soil-borne disease. There are no visible symptoms in the leaves in the early onset stage, so it is not easy to monitor from the plant surface. It cannot be recognized by ordinary spectral images or RGB images of oilseed rape leaves. In this study, hyperspectral imaging is used as monitoring technology, combined with deep learning to build an early identification model of sclerotinia stem rot on oilseed rape to solve the problem of early identification of sclerotinia stem rot on oilseed rape. In this study, the stem rot on oilseed rape was used as the research object, and the mycelium inoculation method was used to induce the disease in the root of oilseed rape. The hyperspectral images of diseased rape plants and healthy plants were collected on the 2nd, 5th, 7th and 9th day after onset. After removing the background, S-G smoothing of the spectral curve, cutting and segmentation, the model training and testing dataset was constructed. Based on the resnet50, the number of feature images was improved, and the first layer’s convolution kernelsize was reduced to improve the model’s recognition ability. The model’s recognition performance and generalization ability were verified based on cross validation. The accuracy of the three models with different structures was 66.79%, 83.78% and 88.66% respectively. The accuracy of the improved model was increased by 16.99% and 4.88% respectively, and the precision and recall rate were improved too. The average accuracy of the improved resnet50 model was 88.66%, the precision and recall rate was more than 83%, and only the recall rate on the seventh day of onset was 79.04%. If the model is binary whether the rape is under disease stress, the accuracy of the model is 97.97%, the precision is 99.19%, and the recall rate is 98.02%. At the same time, the accuracy of the model for the test dataset reached 91.25%.The results of cross-validation showed that the improved model had a good recognition ability for sclerotinia stem rot on oilseed rape within one week and could be used to identify the different stages of sclerotinia stem rot on oilseed rape. The improved model has a stronger ability to identify whether rape was stressed by sclerotinia stem rot on oilseed rape, and the accuracy, precision and recall rate all reached more than 97.97%. At the same time, the model’s accuracy for the test dataset(Day 9 of onset) reached 91.25%, indicating that the model had a good generalization ability for the early recognition of sclerotinia stem rot on oilseed rape. This study solved the problem that asymptomatic disease recognition cannot be carried out based on GRB images and provided are ference for the development of crop diseases early recognition.
梁万杰，冯 辉，江 东，张文宇，曹 静，曹宏鑫. 高光谱图像结合深度学习的油菜菌核病早期识别[J]. 光谱学与光谱分析, 2023, 43(07): 2220-2225.
LIANG Wan-jie, FENG Hui, JIANG Dong, ZHANG Wen-yu, CAO Jing, CAO Hong-xin. Early Recognition of Sclerotinia Stem Rot on Oilseed Rape by Hyperspectral Imaging Combined With Deep Learning. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2023, 43(07): 2220-2225.
[1] DAI Xiao-feng, YE Zhi-hua, CAO Ya-zhong, et al(戴小枫, 叶志华, 曹雅忠, 等). Chinese Journal of Applied Ecology(应用生态学报), 1999, 10(1): 119.
[2] Carvajal-Yepes M, Cardwell K, Nelson A, et al. Science, 2019, 364(6447): 1237.
[3] ZHANG Ning, YANG Gui-jun, ZHAO Chun-jiang, et al(张 凝, 杨贵军, 赵春江, 等). National Remote Sensing Bulletin(遥感学报), 2021, 25(1): 403.
[4] Zhang Chu, Feng Xuping, Wang Jian, et al. Plant Methods, 2017, 13: 39.
[5] Liu F, Shen T, Wang J, et al. Transactions of the ASABE, 2019, 62(1): 123.
[6] Oerke E C, Herzog K, Toepfer R. Journal of Experimental Botany, 2016, 67(18): 5529.
[7] GUI Jiang-sheng, WU Zi-xian，LI Kai(桂江生, 吴子娴, 李 凯). Journal of Zhejiang University(Agric. & Life Sci.)[浙江大学学报（农业与生命科学版)], 2019, 45(2): 256.
[8] HUANG Wen-jiang, SHI Yue, DONG Ying-ying, et al(黄文江, 师 越, 董莹莹, 等). Smart Agriculture(智慧农业), 2019, 1(4): 1.
[9] ZHOU Hong-ping, HU Yi-lei, JIANG Hong-zhe, et al(周宏平, 胡逸磊, 姜洪喆, 等). Transactions of the Chinese Society for Agricultural Machinery(农业机械学报), 2021, 52(5): 308.
[10] Kong Wenwen, Zhang Chu, Cao Feng, et al. Sensors, 2018, 18: 1764.
[11] Kong Wenwen, Zhang Chu, Huang Weihao, et al. Sensors, 2018, 18: 123.
[12] YANG Yan, HE Yong(杨 燕, 何 勇). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2013, 29(20): 135.
[13] LI Bo, LIU Zhan-yu, HUANG Jing-feng, et al(李 波, 刘占宇, 黄敬峰, 等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2009, 25(9): 143.
[14] Hinton G E，Salakhutdinov R R. Science, 2006, 313: 504.
[15] WANG Xian-feng, ZHANG Chuan-lei, ZHANG Shan-wen, et al(王献锋, 张传雷, 张善文, 等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2018, 34(14): 157.
[16] SONG Yu-qing, XIE Xi, LIU Zhe, et al(宋余庆, 谢 熹, 刘 哲, 等). Transactions of the Chinese Society for Agricultural Machinery(农业机械学报), 2020, 51(8): 196.
[17] ZHAO Li-xin, HOU Fa-dong, LÜ Zheng-chao, et al(赵立新, 侯发东, 吕正超, 等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2020, 36(7): 184.
[18] YAO Qing, GU Jia-le, LÜ Jun, et al(姚 青, 谷嘉乐, 吕 军, 等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2020, 36(15): 182.
[19] CHEN Xin-xin, LIU Zi-yi, LÜ Mei-qiao, et al(陈欣欣, 刘子毅, 吕美巧, 等). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2019, 39(3): 730.
