Vibe cartography sandbox

David Oesch has open-sourced his workflow for generating #AI-driven #topographicmaps directly from #aerialimagery, following his earlier experiment in “vibe cartography”. The GitHub repository lets anyone reproduce the method using Python and a Google account, while also inviting collaboration on the proposed future improvements.
Author

Ralph Straumann

Published

October 13, 2025

Roughly two weeks ago, David Oesch (previously) published an experiment in vibe cartography. Now, David has open-sourced his approach for directly producing a topographic map from aerial imagery using AI1 and his experimental results on GitHub.

The repository enables anybody to reproduce David’s approach (using Python 3.8+ and a Google account) and to maybe collaborate towards the “Future improvements” outlined.

Some trial applications of “AI Topographic Maps” by David Oesch

Footnotes

  1. Specifically, a large language model (LLM) and a large multimodal model (LMM).↩︎