Soybean seed vigor discrimination by using infrared spectroscopy and machine learning algorithms

Abstract

A novel approach to distinguish soybean seed vigor based on Fourier transform infrared spectroscopy (FTIR) associated with chemometric methods is presented. Batches with high and low vigor soybean seeds were analyzed. Support vector machine (SVM), K-nearest neighbors (KNN), and discriminant analysis were applied to the raw spectral and reduced-dimensionality data from PCA (principal component analysis). Proteins, fatty acids, and amides were identified as the main molecules responsible for the discrimination of the batches. The cross-validation tests pointed out that high vigor soybean seeds were successfully discriminated from low vigor ones with an accuracy of 100%. These findings indicate FTIR spectroscopy associated with multivariate analysis as a new alternative approach to discriminate seed vigor.