Meta-Album FLW Extended: 8,189 Images of 102 Flower Species from Oxford

8,189 images of flowers across 102 categories, originally collected by the Visual Geometry Group at Oxford University in 2008. The Meta-Album version preprocesses the images by cropping, squaring, and resizing them to 128x128 pixels with an anti-aliasing filter. The dataset features flowers commonly found in the UK, with large variations in scale, pose, and lighting.

Use Cases

• Fine-grained image classification based on 102 distinct flower categories.
• Benchmarking few-shot learning algorithms on a standardized, preprocessed image dataset.
• Training models to handle visual variations in scale, pose, and lighting as described in the source.
• Research in automated plant recognition and ecological monitoring using image data.

Strengths

• Contains 8,189 images, providing a substantial volume for model training.
• Covers 102 distinct flower classes, enabling fine-grained classification tasks.
• Images are preprocessed to a uniform 128x128 size, ensuring consistency for model input.

Limitations

• The original data is from 2008, which may limit relevance for contemporary ecological studies.
• Description metadata is limited; actual data quality and class balance require manual inspection after download.
• Geographic coverage is focused on flowers commonly found in the UK, which may introduce regional bias.

Provenance

Source

Visual Geometry Group, University of Oxford, England.

Collection Method

Images were gathered from different websites and photographed by the original creators, then preprocessed by Meta-Album.

Time Range

Original dataset created in 2008.

Freshness

The Meta-Album version was created on 01 March 2022, but the underlying source data is from 2008.

Geography

Flowers commonly found in the United Kingdom.