Identification and characterization of crops through the analysis of spectral data with machine learning algorithms

NICOLAS RIGALLI; ENRIQUE MONTERO BULACIO; ROMAGNOLI, M.; LUCAS TERISSI; MARGARITA PORTAPILA

CABA

Simposio; 10 Congreso de Agroinformática (47 JAIIO); 2018

This paper assesses the capability of an spectrometer used in field experiments of soybean, maize and wheat. The objective of this work is to select different wavelengths intervals of the spectral reflectance curve, within the range 650-1100nm, as features for classification using machine learning methods. Two different classifications are presented, species selection and growth stage identification. For species classification accuracy of 93% is reached, while 98% is obtained for stage classification. In addition we propose a new index that outperforms analyzed established vegetation indices, which shows the potential advantage of using this type of devices.