Transfer Learning Approach to Seed Taxonomy: A Wild Plant Case Study

Plant taxonomy is the scientific study of classification and naming various plant species. It a branch biology that aims to categorize organize diverse variety life on earth. Traditionally, has been performed using morphological anatomical characteristics, such as leaf shape, flower structure, seed fruit characters. Artificial intelligence (AI), machine learning, especially deep learning can also play an instrumental role in by automating process categorizing species based available features. This investigated transfer techniques analyze images plants extract features be used cluster hierarchically k-means clustering algorithm. Several pretrained models were employed evaluated. In this regard, two separate datasets comprising wild collected from Egypt. Extensive experiments method (DenseNet201) demonstrated proposed methods achieved superior accuracy compared traditional with highest 93% F1-score area under curve (AUC) 95%, respectively. That considerable contrast state-of-the-art approaches literature.