Medicinal Plant Identification Using Deep Learning Model

Medicinal plants play a vital role in traditional Indian healthcare, cultivated and harvested for centuries to support wellness and cure ailments. Indian forests, rich in biodiversity, are home to numerous medicinal herbs that sustain ecosystems and benefit society. Given their importance, these plants have historically been a subject of intensive study, yet their accurate identification remains challenging. Identifying herb species, often with subtle visual differences, requires expertise and can be time-consuming, creating a need for more efficient methods. This study proposes a vision-based, deep learning (DL) model to streamline the identification process, enabling rapid and accurate recognition of herb plants. Focusing on six commonly used plants— betel, curry, Tulsi, mint, neem, and Indian beech— the model was trained on a dataset containing 500 images per plant. The images underwent preprocessing, including resizing and augmentation, to enhance sample diversity. For this task, the Mobile Net DL model was chosen due to its efficiency and suitability for real-time applications. The DL model workflow includes training, validation, and testing, with evaluation based on metrics such as accuracy, precision, and recall. Achieving an accuracy rate of 98.3%, the model demonstrated high reliability distinguishing medicinal plants by their leaves.