Back to briefing
Research Switzerland

swiss-ai/nasa (dataset)

The dataset is a public-domain subset of the NASA Image and Video Library, filtered and captioned with a Qwen vision-language model. It contains 36,997 images covering spaceflight, astronomy, planetary science, Earth observation and aeronautics from NASA centers such as JPL, KSC, MSFC, JSC and GSFC. The images were sourced directly from NASA's public API and prepared by the Swiss AI Initiative vision team.

Published
13 Aug 2026
2 min read 1 source
Intelligence brief
Switzerland / Research
CH

The dataset is a public-domain subset of the NASA Image and Video Library, filtered and captioned with a Qwen vision-language model. It contains 36,997 images covering spaceflight, astronomy, planetary science, Earth observation and aeronautics from NASA centers such as JPL, KSC, MSFC, JSC and GSFC. The images were sourced directly from NASA's public API and prepared by the Swiss AI Initiative vision team.

Why it matters

It provides Swiss AI researchers with a large, well-captioned image collection for training and evaluating vision-language models.

Primary source record
Hugging Face Hub

Original source record. Open the original record to verify the underlying announcement.