OSMGen: Highly Controllable Satellite Image Synthesis using OpenStreetMap Data
Amir Ziashahabi, Narges Ghasemi, Sajjad Shahabi · 2025
Read OSMGen: Highly Controllable Satellite Image Synthesis using OpenStreetMap Data by Amir Ziashahabi, Narges Ghasemi, Sajjad Shahabi (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-00345.txt.
Description
Amir Ziashahabi and five co-authors introduce OSMGen, a system that creates realistic satellite images straight from OpenStreetMap data. Planners and monitors need current images of specific urban features, but curated datasets of features and their changes are scarce. Unlike earlier tools that start from flat map tiles, OSMGen reads the full map data: shapes, tags, place, and time.
That rich input gives fine control over each scene. Editing the map data produces matched before-and-after image pairs, where only the intended change appears and the rest stays fixed. Such pairs can train detectors, ease class imbalance, and let planners preview projects by editing a map. The paired map-and-image data also points toward a loop where new imagery updates the map itself.
Catalog record
| CREATOR | Amir Ziashahabi, Narges Ghasemi, Sajjad Shahabi |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 29 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-00345