MedLeaf-16: A multiclass image dataset of medicinal leaves for identification in computer vision research

Description

MedLeaf-16 is a multiclass image dataset containing 18,595 labeled images of medicinal plant leaves representing 16 plant species commonly used in herbal and traditional medicine. The dataset was created to support research in computer vision, machine learning, and deep learning–based plant identification systems. Each image corresponds to a single leaf sample and is organized into class-specific directories according to plant species. All images were captured using consumer smartphone cameras, including a Samsung Galaxy M12 (48 MP) and a Mi 11X smartphone equipped with an 8 MP wide camera (f/1.8 aperture, 26 mm focal length) based on the Sony IMX582 sensor. Data acquisition was conducted in a controlled indoor environment, where leaves were placed on a white paper background under artificial lighting conditions to maintain consistent image quality. Images were resized to a uniform resolution of 960 × 1280 pixels. Leaf samples were collected from four geographic locations in Bangladesh: Jashore, Faridpur, Savar, and Naogaon. The dataset includes variations in leaf orientation, shape, color, and texture while maintaining consistent capture conditions. Plant species identification was verified by an expert from Sher-e-Bangla Agricultural University, Dhaka, Bangladesh. The MedLeaf-16 dataset can be used for applications such as image classification, feature extraction, transfer learning, and automated medicinal plant recognition in agricultural and computer vision research.

Institutions

• Daffodil International University

  Dhaka Division, Dhaka
• Southeast University

  Dhaka Division, Dhaka

Categories

Licence