Meta Album Flowers Mini: 102 Flower Classes for Few-Shot Learning

4080 images of flowers collected in 2008 by researchers at Oxford University. The dataset contains 102 classes of flowers commonly found in the UK, with images showing variations in scale, pose, and lighting. This version is preprocessed by Meta-Album, with images cropped and resized to 128x128 pixels.

Use Cases

• Benchmarking few-shot learning algorithms based on the 102 flower classes.
• Training image classifiers for plant recognition based on the described variations in scale, pose, and lighting.
• Developing meta-learning models for transfer across visual domains using the preprocessed 128x128 image format.

Strengths

• Contains 4080 images across 102 distinct flower classes.
• Images are preprocessed to a uniform 128x128 pixel size, which may simplify model input pipelines.
• Originates from a well-known source (Visual Geometry Group, Oxford) with a cited academic publication.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Last update date is unknown; freshness unverified.
• Data may reflect geographic bias inherent to the source, as the flowers are described as commonly found in the UK.

Provenance

Source

Visual Geometry Group, University of Oxford

Collection Method

Gathered from different websites and photographed by the original creators.

Time Range

2008

Geography

United Kingdom