Qwen3-VL-Embedding-2B Used to Recreate NASA's 'Your Name in Landsat' in a Weekend

Key Info

A Qwen developer recreated NASA's "Your Name in Landsat" project over a single weekend using Qwen3-VL-Embedding-2B, searching 20 million satellite image patches for letters formed by natural features.

Highlights

• The project uses Qwen3-VL-Embedding-2B as the retrieval model over 20 million satellite image patches.
• It finds letters hidden in natural terrain, mirroring NASA's "Your Name in Landsat" concept.
• The developer completed the build in just a weekend.
• The project is credited to community developer @VoyagerXxxxx, with a link shared by Qwen Developers.