Recognizing rice diseases with modern computer vision techniques

 Relevance. Today, in the fight against rice diseases, it is still widely practiced to uniformly spray the entire field, either as a preventive measure or when any symptoms of disease are detected. Moreover, diseases in the early stages are often identified incorrectly and the complex of drugs is  selected incorrectly. The article explores the possibility of detecting and  classifying some fungal rice diseases from photography using machine learning. Two diseases are considered: blast disease and a group of diseases 
– brown spot.

Methodology. The main idea behind convolutional neural networks is to
try to bring the network closer to how human vision works. To determine the presence of a particular disease in the image, modern computer vision methods based on convolutional neural networks are used. The collection of a dataset must first of all be oriented towards the end user of the model. But even keeping an eye on the quality and shooting conditions both when collecting data and when using a trained model, a number of problems of a fundamental nature can arise that can significantly degrade the quality of the model. Among them: insufficient sample size; natural invariance of predictions with respect to rotations/reflections of the image; instability of predictions, when even insignificant noise can change the result; the effect of overfitting, when the quality of predictions on new images turns out to be significantly lower than on training images. A comparison is made of the four most successful and compact convolutional neural network architectures: GoogleNet, ResNet-18, SqueezeNet-1.0, and DenseNet-121. It was shown that in the data set used for the analysis, the disease can be detected with an accuracy of at least 95%.

Results. The results obtained can be used for automatic recognition of fungal diseases of rice and making a decision on the implementation of protective measures, which could be carried out with minimal labor and time. 

1. Lukyanova IV Varietal features of the stability of rice stems to lodging taking into account their physical and mechanical properties: author. dis ... cand. biol. Sciences / I.V. Lukyanova. – Krasnodar: KubGAU, 2000. – 24 p.

2. Barbedo, J.G.A. A review on the main challenges in automatic plant disease identification based on visible range images // Bio systems Engineering, 144, pp. 52–60. 2016.

3. Vimal K. Shrivastava, Monoj K. Pradhan, Sonajharia Minz, Mahesh P. Thakur. Rice plant disease classification using transfer learning of deep convolutional neural network // The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLII-3/W6, New Delhi, India. 2019.

4. Jitesh P. Shah, Harshadkumar B. Prajapati, Vipul K. Dabhi. A survey on detection and classification of rice plant diseases // In current trends in Advanced Computing (ICCTAC), IEEE International Conference, pp. 1–8. 2016.

5. Pantazi, X.E., Tamouridou, A.A., Alexandridis, T.K., Lagopodi, A.L., Kontouris, G., Moshou, D. Detection of Silybum marianum infection with Microbotryum silybum using VNIR field spectroscopy // Comput. Electron. Agric., 137, 130–137. 2017.

6. Sanyal, P., Patel, S.C. Pattern recognition method to detect two diseases in rice plants // The Imaging Science Journal, 56, pp. 319–325. 2008.

7. Joshi, A.A., Jadhav, B.D. Monitoring and controlling rice diseases using image processing techniques. International Conference on Computing, Analytics and Security Trends (CAST), pp. 471–476. 2016.

8. Bidaux J.M. Screening for horizontal resistance to rice blast (Pyricularia oryzae) in Africa // In: Buddenhagen I.W., Persley G.J. (Eds.). Rice in Africa. London: Acad. Press, pp.159–174. 1978.

9. Methodological guidelines for the diagnosis, accounting and assessment of the harmfulness of rice blast disease // VNIIF / Otv. ed. Kiryukhina R.I.M .: VASKHNIL, 1988. – 40 p.

10. Aneja, K.R. Experiments in Microbiology Plant Pathology and Biotechnology // 4th edn New Age International Publishers, New Delhi. 2005.

11. Jena K.K., Moon H.P., Mackill D.J. Marker assisted selection – a new paradigm in plant breeding // Korean J. Breed., V.35. P. 133–140. 2003.

12. Mukhina Zh.M. Creation of intragenic molecular markers of rice to increase the efficiency of breeding and seed-growing processes / Zh.M. Mukhina, S.V. Tokmakov, Yu.A. Myagkikh, E.V. Dubina // Polythematic network electronic scientific journal of the Kuban State Agrarian University (Scientific journal KubGAU) [Electronic resource]. – Krasnodar: KubGAU, 2011. – No. 03 (67). – Informregister code: 0421100012 0115 - Access mode: http://ej.kubagro.ru/2011/03/pdf/19.pdf.

13. Huy Do Rice Diseases Image Dataset: An image dataset for rice and its diseases from: https://www.kaggle.com/minhhuy2810/rice-diseases-image-dataset.2019.