Recognition of Soybean Diseases Using Machine Learning Techniques Based on Segmentation of Images Captured By UAVs

• Gercina da Silva UFMS
• Alessandro Ferreira UFMS
• Denilson Guilherme Universidade Católica Dom Bosco
• José Fernando Grigolli Fundação MS
• Vanessa Weber UEMS
• Hemerson Pistori Universidade Católica Dom Bosco

DOI: https://doi.org/10.5753/wvc.2020.13476

Resumo

Soybean is an important product for the Brazilian economy, however it has factors that can limit its productive income, like the diseases that are generally difficult to control. Thus, this article aims to use a computer program to recognize diseases in images obtained by a UAV in a soybean plantation. The program is based on computer vision and machine learning, using the SLIC algorithm to segment the images into superpixels. To achieve the objective, after the segmentation of the images, an image dataset was created with the following classes: mildew, target spot, Asian rust, soil, straw and healthy leaves, totaling 22,140 images. Diagrammatic scales were used to assess disease severity. The disease recognition computer program explored four supervised learning techniques: SVM, J48, Random Forest and KNN. The techniques that obtained the best performance were SVM and Random Forests, taking into account the results obtained with all the evaluation metrics used. It was found that the program is efficient to differentiate the classes of diseases treated in this article.

Palavras-chave: soybean diseases, segmentation, UAVs

Referências

EMBRAPA - EMPRESA BRASILEIRA DE PESQUISA AGROPECUÁRIA - EMBRAPA SOJA. Tecnologias de produção de soja – região central do Brasil 2014. - Londrina: Embrapa Soja, 2014.

COMPANHIA NACIONAL DE ABASTECIMENTO – CONAB. Acompanhamento da Safra Brasileira de Grãos. V7. Safra 2019/2020. Brasília. Set/2020. Disponível em <https://www.conab.gov.br/info-agro/safras/graos/boletim-da-safra-de-graos>. Acesso em: 12 de setembro de 2020.

J. F. J. GRIGOLLI, Manejo de Pragas e Doenças na Cultura da Soja. Palestra: Fitossanidade Safra 2015. Fundação MS. Disponível em <http://migre.me/rdAYC> Acesso em Ago/2015. [4] M. J. AFRIDI; X. LIU; J. MITCHELL MCGRATH. "An Automated System for Plant-level Disease Rating in Real Fields," in Proceedings of the 22nd International Conference on Pattern Recognition (ICPR 2014), Stockholm, Sweden, August 24-28, 2014.

M. S. BONALDO; I. C. RIEDO; A. R. LIMA. Monitoramento e diagnóstico de doenças foliares da cultura da soja na região COMCAM na safra 2007/2008. Campo Digital, Campo Mourão, v. 4, n. 1, p. 127–136, 2009.

Y. JIA; Z. SU; W. SHEN; J. YUAN; Z. XU. UAV Technology and Its Application in Agriculture. Advanced Science and Technology Letters. Vol.137 (SUComS 2016), pp.107-111 http://dx.doi.org/10.14257/astl.2016.137.20

J. M. PEÑA-BARRAGAN; K. M de CASTRO; F. LOPEZ-GRANADOS. Object-based approach for crow row characterization in UAV images for site-specific weed management. In Queiroz-Feitosa et al.., editors. 4th International; Conference on Geographic Object-Based Image Analysis (GEOBIA 2012), Rio de Janeiro, Brazil: 426-430

D. GÓMEZ-CANDÓN, A.I.D CASTRO; F. LÓPEZ-GRANADOS. Assessing the accuracy of mosaics from unmanned aerial vehicle (UAV) imagery for precision agriculture purposes in wheat. Precis. Agric. 2014-15, 44–56.

V. UGALE; D. GUPTA. A Comprehensive Survey on Agricultural Image Processing. International Journal of Science and Research (IJSR). Volume 5 Issue 1, January 2016

W. BUSSAB; P. A. MORETTIN. (2011). Estatística básica (7a ed.). São Paulo: Saraiva.

X. REN; J. MALIK. Learning a classification model for segmentation. IEEE ICCV, pp. 10–17, 2003.

R. ACHANTA; K. SMITH; A. LUCCHI; P. FUA; S. SUSSTRUNK. SLIC superpixels. Technical report, EPFL, Tech.Rep. 149300, 2010.

J. LV. An Improved SLIC Superpixels using Reciprocal Nearest Neighbor Clustering. International Journal of Signal Processing, Image Processing and Pattern Recognition Vol. 8, No. 5 (2015), pp. 239-248. Disponível em: <http://dx.doi.org/10.14257/ijsip.2015.8.5.25> Acesso em 20 de jul. de 2015

B. LANTZ. Machine Learning with R. Packt, 2° Edição, 2015. [15] W. T. ANDRADE; L. N. B. QUINTA; A. B. GONCALVES; M. P. CEREDA; H. PISTORI. Segmentação Baseada em Textura e Watershed aplicada a Imagens de Pólen. In: SIBGRAPI 2012 - Conference on Graphics, Patterns and Images, Workshop of Undergraduate Work (WUW), 2012, Ouro Preto - MG. Anais do SIBGRAPI, 2012.

H. ABDI; L. WILLIAMS, L. Newman-Keuls; and Tukey test. In Salkind, N., Frey, B., & Dougherty, D. (Eds.), Encyclopedia of Research Design, pp. 897–904. Sage, Thousand Oaks, CA, (2010).

EMBRAPA SOJA. Tecnologias de produção de soja: região Central do Brasil, 2003. Londrina, 2004. 239p.

G. G. SILVA. Superpixel e aprendizagem supervisionada para a identificação de doenças da soja em imagens obtidas por veículos aéreos não tripulados - Tese (doutorado em Ciências Ambientais e Sustentabilidade Agropecuária) – Universidade Católica Dom Bosco, Campo Grande -MS. 114 p. 2017.

E. C. TETILA, B. B. MACHADO, G. K. MENEZES, A. OLIVEIRA, M. A. VEGA, W. P. AMORIM, N. A. S. BELETE, G. G. SILVA; H. PISTORI. Automatic recognition of soybean leaf diseases using UAV images and deep convolutional neural networks, IEEE Geoscience and Remote Sensing Letters, 2019.