Synthetic OCR Vietnamese — license summary
==========================================

Files: clean/*.png, noisy/*.png, ground_truth.jsonl

License: CC0 1.0 Universal (public domain dedication)

Source text: benchmarks/data/diacritic_eval_v0.txt (also CC0, Neural Research
Lab, 2026-04-25). Rendering script (render.py) is part of nom-vn (Apache 2.0).

Use, copy, modify, redistribute, sell, sublicense — no restrictions, no
attribution required (though it's appreciated).

Full text: https://creativecommons.org/publicdomain/zero/1.0/

Note: The system fonts used by render.py (DejaVu, Lato, FreeSans/FreeSerif)
are NOT redistributed in this folder. They are installed system-wide on most
Linux distros and licensed separately (Bitstream Vera, OFL, GPL respectively).
