Meta-Album RSD: Remote Sensing Scene Images for Few-Shot Learning

800 satellite and aerial images across 20 land cover and scene categories, resized to 128x128 pixels. This dataset is a preprocessed subset of the larger RSD46-WHU collection, created by the Meta-Album project for few-shot learning benchmarks. The original data was collected from Google Earth and Tianditu by researchers at Wuhan University.

Use Cases

• Benchmarking few-shot image classification models based on the 20 scene categories.
• Training models for remote sensing scene recognition based on satellite and aerial imagery.
• Developing lightweight computer vision models based on the standardized 128x128 pixel image size.

Strengths

• Contains 800 images across 20 distinct classes, providing a structured testbed for few-shot learning.
• Images are preprocessed to a consistent 128x128 pixel resolution, simplifying model input pipelines.
• Clear provenance linking to the original RSD46-WHU dataset and associated research papers.

Limitations

• The dataset is relatively small with 800 images, which may limit its use for training large models from scratch.
• Description metadata is limited; actual data quality and class balance require manual inspection after download.

Provenance

Source

RSD46-WHU dataset, collected from Google Earth and Tianditu by researchers at Wuhan University.

Collection Method

Original images were collected from satellite and aerial platforms and resized to 128x128 pixels for Meta-Album.