BMRS: Bongard–Maximov Problems for Remote Sensing

BMRS is a dataset of Bongard–Maximov problems for remote sensing, published on Preprints.org in June 2026. The dataset is authored by Nikita Firsov, Olga Terekhova, and colleagues. It was last updated on the Hugging Face platform on 2026-06-22.

Use Cases

• Benchmarking visual reasoning models based on the Bongard–Maximov problem framework described.
• Training models for remote sensing image classification based on conceptual differences between image sets.
• Developing few-shot or meta-learning algorithms for geospatial imagery based on the problem structure.
• Studying the abstraction capabilities of computer vision systems on remote sensing data.

Strengths

• Associated with a formal preprint publication (doi:10.20944/preprints202606.1484.v1).
• Has a specific, well-defined purpose for remote sensing and visual reasoning.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file size are unknown, which may limit suitability assessment.

Provenance

Source

Preprints.org (doi:10.20944/preprints202606.1484.v1)

Freshness

Last updated 2026-06-22 11:36:56; freshness should be verified.