Identificação das variedades crioulas de feijão-caupi a partir de imagens digitais de sementes e técnicas de inteligência artificial.

Abstract

Cowpea is a crop of great importance worldwide, which is why many landraces and improved cultivars are exploited. The seeds of these materials have intrinsic characteristics that distinguish them. Thus, the objective of this work was to adjust the machine learning models for cowpea identification from the processing of digital images of seeds using artificial intelligence techniques. For that, digital images of seeds of 6 landrace varieties and 10 cultivars were obtained and processed using the vectors InceptionV3, SqueezeNet, VGG16 and VGG19. Subsequently, the Machine learning algorithms K-Nearest Neighbors (KNN - number of nearest neighbors), Decision Tree (Tree), Random Forest (RF - Random Forest), Gradient Boosting (GB - Gradient Boosting), Support Vector Machines (SVM - Support Vector Machines) and Artificial Neural Network (MLP - Multi-Layer Perceptron). The best performance indicator for cowpea identification from digital seed image processing was obtained using the Artificial Neural Network machine learning algorithm.