Machine Learning Classifiers for Distinguishing Various Plant Families Based on Feature Extraction Using Real Data

Taxonomical plant classification is an important task in the field of botany, with applications in biodiversity analysis and conservation efforts. In this paper, we present a machine learning approach to taxonomic plant classification based on taxonomy features extracted from a herbarium dataset (Adhikari, in WII herbarium dataset, https://doi.org/10.15468/dhouv6 [1]) by Wildlife Institute of India. The dataset consists of 4591 occurrences of herbarium plant including their taxon match and geographical distribution. We compare the performance of multiple classifier algorithms on the extracted features and evaluate the accuracy of our model using a variety of metrics. The extracted features are based on the plant hierarchy. Features like phylum, class, order, family, genus, and scientific name are highly correlated and follow the hierarchy order. These primary findings in the dataset helped further for the classification task. The various machine learning classifier algorithms used here are logistic regression, K-nearest neighbor (KNN), support vector machine (SVM), decision tree, and random forest algorithm. Our results show that our approach significantly improves the accuracy of plant classification using machine learning algorithms like decision tree and their geographical occurrence compared to traditional methods and holds promise for future applications in the herbarium plant research field and also demonstrates the potential for using machine learning to aid in the study of plant biodiversity.