Classificação de sementes aquáticas do Pantanal Brasileiro utilizando árvores de decisão

This article compares methods of classifying aquatic seeds.These seeds are taken from seed banks of the Brazilian Pantanal.Twelve species were selected, each with 20 samples.The data bank has 240 images.Those images are processed into a vector with 226 attributes.The proposed methods are Naive-Bayes, Logistic Regression, Decision Tree, Extra-Trees, Multiclass AdaBoost, Histogram-based Gradient Boosting and Ensable Extra-Trees.Among these algorithms, the Ensable Extra-Trees achieved the higher accuracy score, reaching 98.12%.