Classificação de Folhas Nativas do Cerrado Usando Aprendizado de Máquina

Edilaine Gonçalves Costa Faria, Deborah Bambil, Daniel Café, Francisco Assis de Oliveira Nascimento

DOI: 10.14209/sbrt.2022.1570810919

Evento: XL Simpósio Brasileiro de Telecomunicações e Processamento de Sinais (SBrT2022)

Keywords: Aprendizagem de máquina Classificadores de folhas Cerrado

Abstract

This work compares 8 machine learning algorithms in terms of their average accuracy in classifying 30 plant species. The classification system has a proprietary implementation and uses leaf photos as input to the classification system. The results show that the Linear Discriminant Analysis algorithm proved to be the most appropriate for this specific problem, reaching an average accuracy of 98.46%.

Download