Identificação de Huanglongbing (HLB) em plantações de citros utilizando redes convolucionais profundas

Miguel N. Marques; Cristiano O. Pontelli; Ely C. de Paiva

doi:10.20906/CBA2022/3507

Miguel N. Marques Faculdade de Engenharia Mecânica (FEM) - UNICAMP, Campinas, SP 13083-860
Cristiano O. Pontelli Máquinas Agrícolas Jacto, Pompeia, SP 17580-000 e Universidade de Marília (UNIMAR), Avenida Hygino Muzzy Filho, 1001, Marília-SP, Cep 17.525–902
Ely C. de Paiva Faculdade de Engenharia Mecânica (FEM) - UNICAMP, Campinas, SP 13083-860

DOI: https://doi.org/10.20906/CBA2022/3507

Keywords: computer vision, DCNN, HLB, deep learning, citrus, inceptionv3

Abstract

The most common method for the identification of HLB infected plants is visual inspection. The present work aims to present a technique using deep convolutional neural networks to identify the presence of the disease in citrus plants using digital camera images. Two types of images are used: leaves with real background (field) and homogeneous background (studio) from two different classes: healthy and sick.