Multi-Class Medicinal Leaf Disease Image Dataset

The availability of diverse and high-quality image datasets is essential for advancing computer vision, deep learning, and explainable artificial intelligence (XAI) for automated medicinal plant disease diagnosis. This dataset comprises images of three medicinal plant species—Centella asiatica, Kalanchoe, and Mikania micrantha—captured under natural field conditions using OnePlus 7T and Apple iPhone 12 smartphones between January 11, 2025, and November 20, 2025. The images were collected from Hajiganj (Chandpur), Ashulia (Birulia), and Uttara (Diyabari), Dhaka, Bangladesh, covering diverse illumination, backgrounds, and viewing conditions. Background removal was performed as a preprocessing step to improve image quality. The original dataset contains 888 RGB images distributed across nine classes: Centella Asiatica Healthy (CAH, 197), Centella Asiatica Insects (CAI, 64), Centella Asiatica Mild Disease (CAM, 78), Kalanchoe Asymptomatic (KAS, 108), Kalanchoe Complex Foliar Damage (KCF, 62), Kalanchoe Symptomatic (KSY, 60), Mikania micrantha Disease Affected (MMA, 87), Mikania micrantha Distorted (MMD, 114), and Mikania micrantha Healthy (MMH, 118).To address class imbalance and establish a standardized benchmark dataset, image augmentation techniques, including rotation, translation, zooming, shearing, horizontal flipping, and brightness adjustment, were applied to increase each class to 1,000 images, resulting in a balanced dataset of 9,000 RGB images. The dataset is intended to support image classification, transfer learning, explainable AI (XAI), and benchmarking of state-of-the-art computer vision and deep learning models for intelligent medicinal plant disease diagnosis.