Meta-Album RSD: Remote Sensing Dataset with 43 Scene Categories

Meta_Album_RSD_Extended is a preprocessed remote sensing image dataset containing 43,825 images across 43 scene categories. The data was originally collected from Google Earth and Tianditu by researchers at WHU and resized to 128x128 pixels for the Meta-Album benchmark. The dataset is released under a CC BY-NC 4.0 license for research and non-profit purposes.

Use Cases

• Few-shot image classification based on 43 distinct remote sensing scene categories.
• Benchmarking computer vision models on satellite and aerial imagery.
• Land cover and land use classification based on scene categories like urban, agricultural, and water bodies.
• Training and evaluating models for remote sensing image analysis tasks.

Strengths

• Contains 43,825 images, providing a substantial volume for model training.
• Images are preprocessed to a consistent 128x128 pixel resolution.
• Covers 43 distinct scene categories relevant to remote sensing analysis.

Limitations

• Row count and original file formats are unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.
• Data may reflect geographic or source bias inherent to the original Google Earth and Tianditu collection.

Provenance

Source

RSD46-WHU (RSIA-LIESMARS-WHU), Google Earth, Tianditu

Collection Method

Collected from satellite and aerial imagery platforms.