MiniImageNet: Stratified Train-Validation Splits for Few-Shot Learning

MiniImageNet is a widely used benchmark dataset for few-shot image classification tasks. The dataset appears to provide stratified splits for training and validation sets, likely derived from the original ImageNet collection. It is hosted on Kaggle, but the original author, specific size, and license details are not provided in the metadata.

Use Cases

• Benchmarking few-shot image classification models (inferred from domain, verify after download)
• Developing meta-learning algorithms for rapid adaptation to new visual classes (inferred from domain, verify after download)
• Comparing the performance of different stratified sampling strategies for validation splits (inferred from domain, verify after download)

Strengths

• Published on Kaggle, a platform with established data sharing infrastructure.
• The title indicates the splits are stratified, which suggests a methodical approach to data partitioning.

Limitations

• Metadata is minimal; actual content requires verification after download.
• Column-level documentation, file formats, dataset size, and license information are unknown.
• The original source, author, and last update date are not specified.

Provenance

Source

Likely derived from the ImageNet dataset, but the specific source is not stated.

Collection Method

Method of creation and stratification is unknown.

Time Range

Temporal coverage is unknown.

Freshness

Last updated date is unknown; freshness unverified.

Geography

Spatial coverage is unknown.