Project-AgML/medicinal_leaf_species_classification

Medicinal Leaf Species Classification

Leaf images of six medicinal plant species photographed in plant nurseries in Savar, Dhaka, Bangladesh: Arjun (Terminalia arjuna), curry leaf (Murraya koenigii), marsh pennywort (Hydrocotyle vulgaris), mint (Mentha spp.), neem (Azadirachta indica) and 'Rubble Leaf' (Kalanchoe pinnata). Plants were identified with expert help and photographed under natural light with an iPhone 13 Pro, and every image was resized to 512 x 512. The default raw config holds the 1,380 original photos (230 per species); the augmented config holds the authors' 9,660 augmented images (1,610 per species), produced with seven transformations including brightness changes, geometric transforms and horizontal flips. The dataset contains raw and augmented versions.
The raw dataset contains 1,380 images.
Images per class:

• Arjun Leaf: 230
• Curry Leaf: 230
• Marsh Pennywort Leaf: 230
• Mint Leaf: 230
• Neem Leaf: 230
• Rubble Leaf: 230

The augmented dataset contains 9,660 images.
Images per class:

• Arjun Leaf: 1,610
• Curry Leaf: 1,610
• Marsh Pennywort Leaf: 1,610
• Mint Leaf: 1,610
• Neem Leaf: 1,610
• Rubble Leaf: 1,610

This dataset is indexed on https://project-agml.github.io/ as part of the AgML python library.

Citation

@article{matin2025extensive,
  title={An extensive visual data for reliable identification of medicinal plant leaves},
  author={Matin, Md. Mafiul Hasan and Sefatullah, Md. and Kabir, Md Shariar and Habib, Haifa Binte},
  journal={Data in Brief},
  volume={63},
  pages={112129},
  year={2025},
  publisher={Elsevier}
}

The dataset itself can be cited as:

Mafi, M. M. H. M. (2023). Medicinal Plant Identification Dataset [Dataset]. Mendeley Data. https://doi.org/10.17632/FJ93RRFV2Y.2

This dataset was reformatted from its original format to match HuggingFace standards.