Título: Use of Convolutional Neural Networks in the Diagnosis of Corn Diseases
Autores: Paulo Lima, Edson Costa, Maria Holanda, Dhian Oliveira, Esley Santo, Luiz Ramos, Lucas Soares, Isadora Santos, Fabricio Araujo, Jakelyne Silva, Gilberto Souza Jr. and Marcus Braga.
Resumo:
The detection of corn (maize) crop diseases is traditionally carried out by farmers, based on their experience accumulated over a period of field practice. However, the visual observation may represent a risk of error due to subjective perception. This article presents an approach based on Deep Learning to identify diseases that affect corn crops. A public database with 3,852 images of maize plant leaves was used, dividedinto four classes: healthy corn, exserohilun leaf spot (northern leaf blight), common corn rust (common rust) and cercosporiosis (cercospora leaf/gray leaf). The proposed model used Convolutional Neural Networks (CNN) techniques for image classification. The four experiments indicated results with an average accuracy above 94.5%. These results in the identification and diagnosis of plant diseases can contribute significantly as atool to the improvement of the production chain that affect corn crops. All data are available at https://github.com/npcaufra/classificacao-doencas-milho .
Palavras-chave:
Corn Diseases, Classification, Data Augmentation, Convolutional Neural Networks.
Páginas: 5
Código DOI: 10.21528/CBIC2021-27
Artigo em pdf: CBIC_2021_paper_27.pdf
Arquivo BibTeX: CBIC_2021_27.bib