A method for counting and classifying aphids using computer vision.

• Autor(a) principal: LINS, E. A.
• Data de Publicação: 2020
• Outros Autores: RODRIGUEZ, J. P. M., SCOLOSKI, S. I., PIVATO, J., LIMA, M. B., FERNANDES, J. M. C., PEREIRA, P. R. V. da S., LAU, D., RIEDER, R.
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Institucional da EMBRAPA (Repository Open Access to Scientific Information from EMBRAPA - Alice)
• Texto Completo: http://www.alice.cnptia.embrapa.br/alice/handle/doc/1119503; https://doi.org/10.1016/j.compag.2019.105200
• topic: Aphids; Counting; Classification; Computer vision; Measurement

Aphids are insects that attack crops and cause damage directly, by consuming the sap of plants, and indirectly,by vectoring microorganisms that can cause diseases. Cereal crops are hosts for many aphid species, includingRhopalosiphumpadi(an economically important aphid species). Recording and classifying aphids are necessaryfor evaluating and predicting crop damage. Thus, serving as a basis for decision making on the utilization ofcontrolmeasures.Itcanalsobeusefultoevaluateplantresistancetoaphids.Traditionally,therecordingprocessis manual and depends on magnification and well-trained staff. The manual counting is also a time-consumingprocess and susceptible to errors. With this in mind, this paper presents a method and software to automate thecounting and classification ofRhopalosiphum padiusing image processing, computer vision, and machinelearningmethods.Thetextalsopresentsacomparisonofmanuallycountsfromexpertsandvaluesobtainedwiththe software, considering 40 samples. The results showed strong positive correlation in counting and classifi-cation(rs=0.92579)and measurement (r=0.9799).Concluding, thesoftware provedtobe reliableandusefulto aphid population monitoring studies.