基于图像的水稻纹枯病智能测报方法

doi:10.3864/j.issn.0578-1752.2022.08.007

摘要/Abstract

摘要：

【目的】目前水稻纹枯病测报依赖人工调查水稻发病丛数、株数和每株严重度来计算其病情指数,操作专业性强,费时费力且数据难以追溯。本研究提出基于图像的水稻纹枯病病斑检测模型和发生危害分级模型,为水稻纹枯病智能测报提供理论依据。【方法】利用便携式图像采集仪采集田间水稻纹枯病图像,研究不同目标检测模型（Cascade R-CNN和RetinaNet）和特征提取网络（VGG-16和ResNet-101）对水稻纹枯病病斑的检测效果,筛选出具有较好检测效果的模型。针对Cascade R-CNN模型检测纹枯病病斑存在漏检现象,根据纹枯病病斑呈现形状不规则、大小和位置多变的复杂情况,对Cascade R-CNN进行改进,添加OHEM结构均衡难易样本,选择边框回归损失函数,通过精准率、漏检率、平均精度和P-R曲线来评价不同模型的检测效果。在改进的Cascade R-CNN-OHEM-GIOU模型检测结果基础上,分别建立基于病斑面积和病斑数的水稻纹枯病丛发生危害分级模型,通过决定系数（R²）和Kappa值筛选分级模型。【结果】在相同主干网络条件下,Cascade R-CNN模型较RetinaNet模型对水稻纹枯病具有更好的检测效果,其中Cascade R-CNN-ResNet-101目标检测模型效果最佳,病斑检测准确率为92.4%,平均精度为88.2%,但漏检率为14.9%。改进的Cascade R-CNN-OHEM-GIOU检测模型有效解决了样本不均衡问题,添加边框回归损失函数有效降低了漏检率,较Cascade R-CNN-ResNet-101模型降低8.7%,平均精度提高到92.3%。以人工分级结果作为标准,基于病斑面积的水稻纹枯病发生危害分级模型在0—5级分级准确率分别为96.0%、90.0%、82.0%、76.0%、74.0%和96.0%,平均分级准确率为85.7%,Kappa系数为0.83,基于图像的水稻纹枯病丛发生危害分级与人工分级结果具有较高的一致性。【结论】基于图像的水稻纹枯病智能测报方法可实现病斑自动检测和发生危害自动分级,提高了测报的智能化水平,结果客观且可追溯,也可为其他农作物病害智能测报提供参考。

关键词: 水稻纹枯病, 病斑图像, 智能测报, Cascade R-CNN模型, 危害分级模型

Abstract:

【Objective】At present, the forecast of rice sheath blight relies on the number of diseased clusters, the number of rice plants and the severity of each plant to calculate the disease index based on manual surveys. The method is highly professional, time-consuming and laborious. The data is difficult to trace. The objective of this study is to propose a detection model of rice sheath blight lesions and a damage grading model of rice sheath blight based on images, and to provide a theoretical basis for the intelligent forecasting of rice sheath blight.【Method】Images of rice sheath blight in paddy field were collected by a portable image acquisition instrument. Different detection models (Cascade R-CNN and RetinaNet) and feature extraction networks (VGG-16 and ResNet-101) were developed to test the detection effect of disease lesions. The best model was chosen. However, the Cascade R-CNN model appeared some missing detection of sheath blight lesions. Because the sheath blight lesions are irregular in shape, and variable in size and location, the Cascade R-CNN model was improved through adding OHEM structure to balance the hard and easy samples in the network and choosing the bounding box regression loss function. The precision rate, missing rate, average precision and P-R curve were used to evaluate the detection effects of different models. Based on the detection results of the improved Cascade R-CNN-OHEM-GIOU model, two damage grading models based on the area and number of disease lesions were developed, respectively. The determination coefficient (R²) and Kappa value were used to choose the damage level model of rice sheath blight. 【Result】Under the same backbone network conditions, the Cascade R-CNN model had a better detection effect on rice sheath blight than the RetinaNet model. The Cascade R-CNN-ResNet-101 model had the best detection effect on sheath blight lesions. The precision rate was 92.4%, the average precision was 88.2% and the missing rate was 14.9%. The improved Cascade R-CNN-OHEM- GIOU model effectively solved the problem of sample imbalance, and effectively reduced the missing rate by adding a border regression loss function, which was 8.7% lower than the missing rate of the Cascade R-CNN-ResNet-101 model, and the average precision was increased to 92.3%. With the results of manual disease grading as the standard, the grading model of rice sheath blight at 0 to 5 grades based on the area of diseased lesions had the accurate rates of 96.0%, 90.0%, 82.0%, 76.0%, 74.0% and 96.0%, respectively. The average grading accuracy rate was 85.7%, and the Kappa coefficient was 0.83. The damage grade results of rice sheath blight based on images were consistent with the manual grading results.【Conclusion】The intelligent forecasting method of rice sheath blight based on images can automatically detect the disease lesions and calculate the damage grade. This method increases the intelligence level and the results are objective and traceable. It may also provide an idea for intelligent forecasting of other crop diseases.

Key words: rice sheath blight, disease lesion image, intelligent forecasting, Cascade R-CNN model, damage grading model

韩晓彤,杨保军,李苏炫,廖福兵,刘淑华,唐健,姚青. 基于图像的水稻纹枯病智能测报方法[J]. 中国农业科学, 2022, 55(8): 1557-1567.

HAN XiaoTong,YANG BaoJun,LI SuXuan,LIAO FuBing,LIU ShuHua,TANG Jian,YAO Qing. Intelligent Forecasting Method of Rice Sheath Blight Based on Images[J]. Scientia Agricultura Sinica, 2022, 55(8): 1557-1567.

0
/ / 推荐

导出引用管理器 EndNote|Reference Manager|ProCite|BibTeX|RefWorks

链接本文: https://www.chinaagrisci.com/CN/10.3864/j.issn.0578-1752.2022.08.007

https://www.chinaagrisci.com/CN/Y2022/V55/I8/1557

图/表 12

图1

图2

图3

图4

表1

图5

表2

图6

图7

图8

图9

表3

参考文献 26

[1]	中华人民共和国国家标准. 稻纹枯病测报技术规范: GB/T 15791- 2011. 北京: 中国标准出版社, (2011-09-29) [2021-11-03].
	National Standards of the People’s Republic of China. Rules of monitoring and forecasting for the rice sheath blight (Rhizoctonia solani Kukn):GB/T 15791-2011. Beijing: China Standards Press, (2011-09-29) [2021-11-03]. (in Chinese)
[2]	王献锋, 张善文, 王震, 张强. 基于叶片图像和环境信息的黄瓜病害识别方法. 农业工程学报, 2014, 30(14): 148-153.
	WANG X F, ZHANG S W, WANG Z, ZHANG Q. Recognition of cucumber diseases based on leaf image and environmental information. Transactions of the Chinese Society of Agricultural Engineering, 2014, 30(14): 148-153. (in Chinese)
[3]	胡耀华, 平学文, 徐明珠, 单卫星, 何勇. 高光谱技术诊断马铃薯叶片晚疫病的研究. 光谱学与光谱分析, 2016, 36(2): 515-519.
	HU Y H, PING X W, XU M Z, SHAN W X, HE Y. Detection of late blight disease on potato leaves using hyperspectral imaging technique. Spectroscopy and Spectral Analysis, 2016, 36(2): 515-519. (in Chinese)
[4]	葛婧. 基于计算机图像处理技术的作物病害等级检测[D]. 合肥: 安徽农业大学, 2007.
	GE J. Detection of crops plant disease rank based on computer image processing technology[D]. Hefei: Anhui Agricultural University, 2007. (in Chinese)
[5]	马德贵, 邵陆寿, 葛婧, 丁克坚, 钱良存. 水稻稻瘟病及水稻纹枯病病害程度图像检测. 中国农学通报, 2008, 24(9): 485-489.
	MA D G, SHAO L S, GE J, DING K J, QIAN L C. Detection of the harm degree of rice blast and rice sheath blight. Chinese Agricultural Science Bulletin, 2008, 24(9): 485-489. (in Chinese)
[6]	袁媛, 陈雷, 吴娜, 李淼. 水稻纹枯病图像识别处理方法研究. 农机化研究, 2016, 38(6): 84-87, 92.
	YUAN Y, CHEN L, WU N, LI M. Recognition of rice sheath blight based on image procession. Journal of Agricultural Mechanization Research, 2016, 38(6): 84-87, 92. (in Chinese)
[7]	KRISHNA R V V, KUMAR S S.Computer vision based identification of nitrogen and potassium deficiency in FCV tobacco//Proceedings of the International Conference on Computational Science and Engineering, 2016: 105-111.
[8]	RAMCHARAN A, BARANOWSKI K, MCCLOSKEY P, AHMED B, LEGG J, HUGHES D P. Deep learning for image-based cassava disease detection. Frontiers in Plant Science, 2017, 8: 1852. doi: 10.3389/fpls.2017.01852
[9]	苏婷婷, 牟少敏, 董萌萍, 时爱菊. 深度迁移学习在花生叶部病害图像识别中的应用. 山东农业大学学报(自然科学版), 2019, 50(5): 865-869.
	SU T T, MU S M, DONG M P, SHI A J. Application of deep transfer learning in image recognition of peanut leaf diseases. Journal of Shandong Agricultural University (Natural Science Edition), 2019, 50(5): 865-869. (in Chinese)
[10]	FUENTES A, YOON S, KIM S C, PARK D S. A robust deep- learning-based detector for real-time tomato plant diseases and pests recognition. Sensors, 2017, 17(9): 2022. doi: 10.3390/s17092022
[11]	RAMESH S, VYDEKI D. Recognition and classification of paddy leaf diseases using optimized deep neural network with Jaya algorithm. Information Processing in Agriculture, 2020, 7(2): 249-260. doi: 10.1016/j.inpa.2019.09.002
[12]	曹英丽, 江凯伦, 于正鑫, 肖文, 刘亚帝. 基于深度卷积神经网络的水稻纹枯病检测识别. 沈阳农业大学学报, 2020, 51(5): 568-575.
	CAO Y L, JIANG K L, YU Z X, XIAO W, LIU Y D. Detection and recognition of rice sheath blight based on deep convolutional neural network. Journal of Shenyang Agricultural University, 2020, 51(5): 568-575. (in Chinese)
[13]	俞佩仕, 郭龙军, 姚青, 杨保军, 唐健, 许渭根, 陈渝阳, 朱旭华, 陈宏明, 张晨光, 段德康, 贝文勇, 彭晴晖. 基于移动终端的稻田飞虱调查方法. 昆虫学报, 2019, 62(5): 615-623.
	YU P S, GUO L J, YAO Q, YANG B J, TANG J, XU W G, CHEN Y Y, ZHU X H, CHEN H M, ZHANG C G, DUAN D K, BEI W Y, PENG Q H. A survey method based on mobile terminal for rice planthoppers in paddy fields. Acta Entomologica Sinica, 2019, 62(5): 615-623. (in Chinese)
[14]	EVERINGHAM M, VAN GOOL L, WILLIAMS C K I, WINN J, ZISSERMAN A. The pascal visual object classes (voc) challenge. International Journal of Computer Vision, 2010, 88(2): 303-338. doi: 10.1007/s11263-009-0275-4
[15]	CAI Z, VASCONCELOS N. Cascade R-CNN: Delving into high quality object detection//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018: 6154-6162.
[16]	REN S, HE K, GIRSHICK R, SUN J. Faster R-CNN: Towards real-time object detection with region proposal networks. Advances in Neural Information Processing Systems, 2015, 28: 91-99.
[17]	SHRIVASTAVA A, GUPTA A, GIRSHICK R. Training region-based object detectors with online hard example mining//Proceeding of the IEEE Conference on Computer Vision and Pattern Recognition, 2016: 761-769.
[18]	REZATOFIGHI H, TSOI N, GWAK J Y, SADEGHIAN A, REID I, SAVARESE S. Generalized intersection over union: A metric and a loss for bounding box regression//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019: 658-666.
[19]	ALTMAN E I, FINANCE M A. Predicting financial distress of companies:Revisiting the Z-Score and ZETA models//Handbook of Research Methods & Applications in Empirical Finance. Edward Elgar Publishing, 2013.
[20]	刘明. 支持向量机中Sigmoid核函数的研究[D]. 西安: 西安电子科技大学, 2009.
	LIU M. The study on sigmoid kernel function in support vector machine[D]. Xi’an: Xidian University, 2009. (in Chinese)
[21]	LIN T Y, GOYAL P, GIRSHICK R, HE K, DOLLAR P. Focal loss for dense object detection//Proceedings of the IEEE International Conference on Computer Vision, 2017: 2980-2988.
[22]	LIU X, CHI M, ZHANG Y F, QIN Y Q. Classifying high resolution remote sensing images by fine-tuned VGG deep networks//IGARSS IEEE International Geoscience and Remote Sensing Symposium, 2018: 7137-7140.
[23]	HE K, ZHANG X, REN S. Deep residual learning for image recognition//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016: 770-778.
[24]	BUCKLAN M, GEY F. The relationship between recall and precision. Journal of the American Society for Information Science, 1994, 45(1): 12-19. doi: 10.1002/(SICI)1097-4571(199401)45:1<12::AID-ASI2>3.0.CO;2-L
[25]	赵松山. 对拟合优度R²的影响因素分析与评价. 东北财经大学学报, 2003(3): 56-58.
	ZHAO S S.Analysis and evaluation of influencing factors of goodness of fit R². Journal of Dongbei University of Finance and Economics, 2003(3): 56-58. (in Chinese)
[26]	许文宁, 王鹏新, 韩萍, 严泰来, 张树誉. Kappa系数在干旱预测模型精度评价中的应用--以关中平原的干旱预测为例. 自然灾害学报, 2011, 20(6): 81-86.
	XU W N, WANG P X, HAN P, YAN T L, ZHANG S Y. Application of Kappa coefficient in accuracy assessments of drought forecasting model: A case study of Guanzhong Plain. Journal of Natural Disasters, 2011, 20(6): 81-86. (in Chinese)

数据集 Dataset	有病斑的图像数 Number of images with lesions	无病斑的图像数 Number of images without lesions	病斑数量 Lesion number	每幅图像平均病斑数 Average lesion number per image
训练集 Training set	4122	896	25093	5.0
验证集 Validation set	1766	384	12257	5.7
测试集 Testing set	250	50	1620	5.4

模型种类 Model type	检测数 Number of detected lesions	正确识别数 Number of correct identified lesions	准确率 Precision rate (%)	漏检率 Missing rate (%)	平均精度 AP (%)
RetinaNet-VGG-16	1404	1230	87.6	24.1	/
RetinaNet-ResNet-101	1444	1311	90.8	19.1	/
Cascade R-CNN-VGG-16	1439	1298	90.2	19.9	/
Cascade R-CNN-ResNet-101	1492	1379	92.4	14.9	88.2
Cascade R-CNN-GIOU	1520	1401	92.2	13.5	89.1
Cascade R-CNN-OHEM-GIOU	1644	1519	92.4	6.2	92.3