Classification of Philippine herbal plants via leaf using different machine learning algorithms
Authors: Robert G. De Luna, Marife A. Rosales, and Elmer P. Dadios
Abstract
Herbal plants have significant role in the field of medicine to cure some known diseases. Presently, the herbal plant identification is performed manually by an expert or person with enough knowledge regarding the plant. Sometimes, manual identification is prone to human error, resulting to incorrect usage of herbal plants. In this study, a machine vison-based herbal plant identification was implemented. An improvised image capturing system with 16 megapixels resolution camera was used with the aid of MATLAB installed in laptop to gather real images of twelve herbal plants. An intelligent system was developed by utilizing image processing, feature extraction, and machine learning (ML) algorithms using Python. The classification accuracy was used to select the best model. Moreover, F1 score metric was used to compare the performances of the default and optimized models in identifying all the herbal classes. Based on the results, SVM model showed the best performance in classifying the herbal plants with accuracy score of 94.50% and 93.30% for the optimized training and testing performances.