MedicinalLeafBD: Medicinal Plant Leaf Image Dataset from Bangladesh for Pharmacological and Agricultural Innovation

The MedicinalLeafBD dataset is a comprehensive collection of high-quality images of medicinal plant leaves, systematically gathered from the National Botanical Garden in Dhaka and various rural areas of Bhola Sadar, Bangladesh, between March 5 and March 15, 2025. The dataset consists of 16 distinct classes representing a diverse range of medicinal plant species commonly used in traditional and herbal medicine practices across Bangladesh. The total dataset contains 6,827 images, with class-wise distribution as follows: Ashok (437 images), Basil (461 images), Cannonball_Tree (424 images), Cordia (426 images), Debdaru (416 images), Heaven_Lotus (406 images), Jarul (414 images), Mastwood (419 images), Minjiri (443 images), Nageshor (419 images), Neem (446 images), Punnag (427 images), Royna (411 images), Telakucha (429 images), Thankuni (433 images), and Vasaka (416 images). The images were captured using the iPhone 15 smartphone in natural lighting conditions, ensuring high resolution and clarity. Later resized to 640 × 480 pixels with a resolution of 72 dpi to standardize the dataset. While the dataset includes image-level class labels, it does not contain detailed annotations such as bounding boxes or segmentation masks. It serves as a valuable resource for researchers and machine learning practitioners focused on medicinal plant species classification and identification. This dataset is particularly relevant for applications in pharmacological research, botanical analysis, and AI-driven plant recognition, where accurate classification of medicinal plants can support healthcare innovation and sustainable agricultural practices.