An extensive visual data for reliable identification of medicinal plant leaves

TL;DR

This paper introduces a large, high-quality dataset of medicinal plant leaf images to help develop automated identification systems using computer vision.

Contribution

The novel contribution is a publicly available dataset of 11,040 medicinal plant leaf images with augmentation for machine vision research.

Findings

The dataset includes 1380 original images of six medicinal plant species.

Seven augmentation techniques generated 9660 additional images for improved model generalization.

The dataset is openly accessible via the Mendeley Data repository.

Abstract

This study presents a high-quality, labeled image dataset of medicinal plant leaves designed to support machine vision and computer vision research. The dataset comprises 1380 original RGB images spanning six medicinal plant species Arjun Leaf, Curry Leaf, Marsh Pennywort Leaf, Mint Leaf, Neem Leaf, and Kalanchoe pinnata (Rubble Leaf) captured under natural lighting conditions using the iPhone 13 Pro. To enhance dataset diversity and improve model generalization, seven augmentation techniques were applied, including brightness adjustment, geometric transformations, and horizontal flipping, resulting in 9660 augmented images and a total of 11,040 samples. All images were pre-processed and resized to 512 × 512 pixels. The dataset was organized into well-structured directories and published openly via the Mendeley Data repository. This dataset offers a valuable resource for researchers developing automated medicinal plant identification systems and contributes to broader efforts in ethnobotany, agriculture, and digital health.

Figures