LAUSR

Dry beans, which are produced in large quantities, have the highest level of genetic diversity. The quality of seeds has a significant impact on crop yield. The importance of seed classification to both marketing and production can be shown by realizing that sustainable agricultural systems depend on these principles. This research is primarily aimed at providing a means to generate uniform seed varieties, as seed is not certified as a single variety. To achieve consistent seed classification, we have proposed Extreme Gradient Boosting ensembles using the Synthetic Minority Over-Sampling Methodology (SMOTE) to differentiate seven distinct registered types of dry beans with similar characteristics. There were a total of 13,611 grains from seven different varieties of dry beans sampled for the classification model. Classification algorithms based on machine learning like Decision Tree (DT), Support Vector Machine (SVM), Multilayer Perceptron (MLP), Adaptive Boosting classifier, Bagging Classifier, and Extreme Gradient Boosting ensembles using the Synthetic Minority Over-Sampling Methodology (SMOTE) were developed and compared. Overall correct classification rates for SVM, MLP, DT, Adaboost, Bagging, and Extreme Gradient Boost classifiers were 94.44%, 94.48%, 96.53%, 96.35%, 96.89%, and 97.32%, respectively. Extreme Gradient Boosting ensembles using the SMOTE classification model have the best accuracy. The results of this study satisfy the producers' and customers' demand for uniform bean varieties.