TAMMI: Multimodal Remote Sensing Visual Question Answering Dataset

TAMMI is a large-scale multimodal Visual Question Answering dataset for remote sensing, introduced at the CVPR 2025 EarthVision Workshop. It combines three complementary satellite image modalities with question-answer annotations automatically generated from official geographic databases. The dataset was authored by HichemBoussaid and last updated on the Hugging Face platform in May 2026.

Use Cases

• Train multimodal VQA models based on the combination of VHR, Sentinel-2, and Sentinel-1 imagery.
• Benchmark remote sensing AI systems on tasks requiring joint understanding of text and multiple image modalities.
• Develop foundation models for geospatial intelligence using automatically generated question-answer pairs from official databases.

Strengths

• Integrates three complementary satellite image modalities: Very High Resolution orthophotos, Sentinel-2 multispectral, and Sentinel-1 SAR.
• Question-answer annotations are automatically generated from official geographic databases, suggesting a structured annotation process.
• Dataset was introduced at a major computer vision conference (CVPR 2025 EarthVision Workshop), indicating peer-reviewed academic context.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for large-scale training.
• Description metadata is limited; actual data quality and coverage require manual inspection after download.

Provenance

Source

Hugging Face dataset authored by HichemBoussaid.

Collection Method

Question-answer annotations were automatically generated from official geographic databases.

Freshness

Last updated 2026-05-29 21:01:52; freshness should be verified.