Meta-Album Sports Actions: 73 Sports Classes with 10,416 Images

Meta_Album_SPT_Extended is a preprocessed version of the 100-Sports image classification dataset, containing 10,416 images across 73 different sports. The dataset was originally compiled by Gerald Piosenka from internet searches and preprocessed for the Meta-Album benchmark by Jilin He in March 2022. Images are resized to 128x128 pixels and are provided under a CC0 1.0 Public Domain license.

Use Cases

• Training few-shot image classifiers based on the 73 distinct sports categories.
• Benchmarking meta-learning algorithms on a standardized, preprocessed image dataset.
• Developing models for sports recognition and action classification.
• Studying domain adaptation for visual tasks within the human actions domain.

Strengths

• Contains 10,416 images, providing a substantial sample for model training and evaluation.
• Covers 73 distinct sports classes, offering a diverse classification challenge.
• Images have been preprocessed to a uniform 128x128 pixel size, ensuring consistency.
• Duplicate images were removed from the original source, which likely improves data quality.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.
• The description does not specify the geographic or temporal distribution of the source images, which may introduce bias.

Provenance

Source

Kaggle (100 Sports Image Classification by Gerald Piosenka), preprocessed for Meta-Album.

Collection Method

Images were gathered from internet searches, duplicates removed, and resized using OpenCV.

Time Range

null

Freshness

Created Date is listed as 01 March 2022; last updated date is unknown.

Geography

null