Year: 2025
Type: article
Abstract: The pharmaceutical, Ayurvedic, and herbal medicine industries depend on the precise identification of therapeutic plants. Botanists must manually inspect plants using traditional methods of identification, which takes time and is prone to mistakes. The use of deep learning (DL) techniques for automated identification of medicinal plants from leaf photos is investigated in this work. Several CNN architectures, such as DenseNet121, MobileNet, and InceptionV3, are compared in order to identify the most effective model for plant classification. Preprocessing methods including scaling and augmentation were employed to improve model performance on a dataset of scanned photos of 50 different types of medicinal plants. Using MobileNet, the suggested system demonstrated its efficacy in real-time plant identification with an accuracy of 98.3%.According to the study, DL models have the ability to help taxonomists, botanists, and the pharmaceutical industry ensure that therapeutic plants are used appropriately.
Authors: Yash Rawat, Aarti Sehwag
Institution: Bharati Vidyapeeth's College of Engineering, Delhi
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4410464319