A dataset of medicinal plant leaves from Bangladesh
Description
This dataset contains a collection of 1,094 original JPG images of medicinal leaves, captured in natural conditions from the rural jungle area of Guadanga village, Phulpur, Mymensingh, Bangladesh. The images are organized into six folders, each representing a distinct and commonly found medicinal plant species: • Centella Asiatica • Coccinia Grandis • Eclipta Prostrata • Mikania Micrantha • Murraya Koenigii • Stephania Japonica The original leaf images were captured using two high-resolution smartphone cameras — Google Pixel 5 and Google Pixel 6a — to ensure clear and detailed visuals suitable for a variety of machine learning and image processing tasks. All images are unprocessed, making the dataset ideal for researchers who wish to work with real-world, raw image data. To support the development of deeper and more robust deep learning models, this dataset also includes 2,188 augmented JPG images, generated from the original set using standard image augmentation techniques. These augmentations help improve generalization and model performance on complex leaf classification tasks. Dataset Summary: Total Images: Original: 1,094 JPG images Augmented: 2,188 JPG images Plant Classes: 6 distinct medicinal plant species Image Format: JPG Image Organization: Organized into 6 folders named after plant species Capture Devices: Google Pixel 5 and Google Pixel 6a Source Location: Jungle of Guadanga, Phulpur, Mymensingh, Bangladesh Purpose: Suitable for tasks such as medicinal plant classification, leaf recognition, image segmentation, and deep learning-based botanical analysis. This dataset provides a rich and realistic foundation for researchers and practitioners working in domains such as plant pathology, botany, herbal medicine identification, and AI-driven agriculture.