Meta-Album Medicinal Leaf: 800 Images of 20 Healthy Herb Species

800 images of healthy medicinal leaves from 20 species, originally captured on a mobile phone and preprocessed to a uniform 128x128 pixel resolution. The dataset was created by Phan Anh VU for the Meta-Album benchmark in March 2022, derived from a source dataset by Roopashree S and Anitha J published on Mendeley Data in 2020.

Use Cases

• Few-shot image classification based on 20 distinct medicinal plant species.
• Benchmarking meta-learning algorithms using a standardized, preprocessed image dataset.
• Training computer vision models for plant identification based on leaf imagery.
• Developing educational tools for recognizing medicinal herbs based on visual features.

Strengths

• Uniform image size of 128x128 pixels, which likely simplifies model input pipelines.
• Includes 800 images across 20 classes, providing a defined scale for few-shot learning tasks.
• Clear licensing under CC BY 4.0 for both the original and Meta-Album releases.
• Images were captured on a uniform white background, which may reduce background noise.

Limitations

• The dataset is a subset of a larger original collection; the description notes the original had around 1,800 images across 30 species, suggesting a reduction in scale and class diversity.
• Image quality and resolution may be limited by the original mobile phone camera capture and subsequent cropping and resizing.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Medicinal Leaf Dataset on Mendeley Data, curated into the Meta-Album benchmark.

Collection Method

Leaves were plucked from plants, placed on a white background, and photographed with a mobile phone. Images were then center-cropped and resized.