Classifying Philippine Medicinal Plants Based on Their Leaves Using Deep Learning

Medicinal plants play an important role in human mental and physical well-being. As a result, recognizing and categorizing medicinal herbs is critical to providing quality medicines. The goal of this study is to create a model for classifying Philippine medicinal plants based on their leaves. The classifier in the study was built using VGG19. The dataset, which contains forty (40) different species of plants and a total of 4922 leaves, yielded an average accuracy of 92.67%. Furthermore, because data is limited, data augmentation is used to reduce computational burden and increase variance in training data. Based on the classifier developed, a prototype application was created to test its accuracy in a real-world scenario. The prototype application correctly identified all 21 leaf species, two out of three leaves from ten classifications, one out of three leaves from eight classifications, and none of three leaves from one category when three (3) leaf samples were tested for each category. According to the findings of the study, the classification of the Philippine medicinal plants could be done using a convolutional neural network using VGG19.