Meta-Album FLW: 800 Images of 20 Flower Species from the UK

800 standardized images of flowers across 20 classes, preprocessed from the original Oxford Flowers dataset. The images were collected from various websites and photographed by the original creators, primarily featuring flowers common in the UK. The dataset was created by Maria-Elena Nilsback and Andrew Zisserman at Oxford University in 2008 and later preprocessed for the Meta-Album benchmark in 2022.

Use Cases

• Benchmarking few-shot image classification models based on the standardized 128x128 pixel format and 20-class structure.
• Training models to recognize plant species based on the dataset's focus on flowers common in the UK.
• Studying model robustness to variations in scale, pose, and lighting mentioned in the original image characteristics.

Strengths

• Contains 800 images across 20 distinct classes, providing a defined scope for few-shot learning tasks.
• Images are preprocessed to a consistent 128x128 pixel size with anti-aliasing, reducing initial data preparation effort.
• Has clear provenance from a well-known academic source (Visual Geometry Group, University of Oxford).

Limitations

• Description metadata is limited; actual data quality, such as label accuracy or image clarity, requires manual inspection after download.
• Last update date for the OpenML entry is unknown; freshness unverified.
• The dataset's geographic focus on flowers common in the UK may introduce regional bias for global classification models.

Provenance

Source

Visual Geometry Group, University of Oxford, England.

Collection Method

Images gathered from different websites and some photographed by the original creators.

Time Range

Original dataset created in 2008.

Freshness

Original data from 2008; Meta-Album preprocessing version created on 01 March 2022.

Geography

Flowers commonly found in the United Kingdom.