Deep Learning-based Automated Identification of Indian Medicinal Plants Using Leaf Imagery

Deep Learning-based Automated Identification of Indian Medicinal Plants Using Leaf Imagery

Dr. Keerthi Kumar HM
Dept. of Computer Science and Engg.

Malnad College of Engineering Hassan, India

Guide

Chirag MR

Dept. of Computer Science and Engg.

Malnad College of Engineering Hassan, India

Gnanesh HJ

Dept. of Computer Science and Engg.

Malnad College of Engineering Hassan, India

Rakshith KV
Dept. of Computer Science and Engg.

Malnad College of Engineering Hassan, India

Bhoomika J

Dept. of Computer Science and Engg.

Malnad College of Engineering Hassan, India

Abstract—The accurate identification of medicinal plants is crucial for ethnobotanical research, pharmaceutical development, and biodiversity conservation. Traditional identification methods rely on expert knowledge, which is time-consuming and not always accessible. This paper proposes a deep learning-based framework using Convolutional Neural Networks (CNNs) for the automated classification of Indian medicinal plants through leaf imagery. We systematically compare the performance of five state-of-the-art CNN architectures: MobileNetV2, ResNet50, InceptionV3, Xception, and DenseNet121. InceptionV3 achieves the best classification accuracy of 95%, demonstrating robust performance in extracting spatial hierarchies of leaf features. Furthermore, we introduce Tinkerit, a lightweight desktop appli- cation with a graphical user interface (GUI), enabling real-time plant recognition. Our solution paves the way for modernizing herbal medicine practices through AI-driven tools, contributing to both healthcare innovation and ecological preservation.