Deep Learning Based Medicinal Plants Identification

Accurate identification of medicinal plants is crucial for applications in healthcare and biodiversity preservation. Traditional methods are labor-intensive and error-prone, necessitating automated solutions. This paper presents a deep learning-based framework utilizing convolutional neural networks (CNNs) for medicinal plant identification. Trained on diverse image datasets, the model achieves high accuracy even under challenging conditions, outperforming traditional machine learning approaches. The model accurately detects the colored leaf images, A damaged or incomplete leaf which overcomes the previous existing models and also increased the count of the number of classes. This work offers a scalable solution to support advancements in ethnobotany and pharmacology. Dataset contains 3027 images of different categories like Crape Jasmine, Hibiscus, Jamun, Mint, Peepal, Pepper, Tulasi. A custom CNN is used in this paper, optimizer Adam, loss function spatial cross entropy, batch size 128 and epochs 150. This model gives the accuracy of 88%.