signa-core

Reusable, GUI-free ArUco marker detection for the Evidentia pipeline.

signa-core is the pure OpenCV/numpy detection layer extracted from the Signa WebODM plugin. It detects ArUco markers in images and returns their ids, corners and centers — leaving the interpretation to the caller:

• Signa (WebODM plugin) matches markers against surveyed coordinates → gcp_list.txt.
• Mensura matches markers against their known physical size → metric scale constraints.

Both share this one detection implementation instead of duplicating it (or depending on find-gcp).

Install / use

pip install signa-core   # or: pip install -e signa-core

from signa_core import detect_markers

# dict_id 1 = DICT_4X4_100 (see DICT_CHOICES)
result = detect_markers(["img1.jpg", "img2.jpg"], dict_id=1, minrate=0.01, ignore=0.33)
# {'img1.jpg': [{'id': 7, 'corners': [[x, y], ...4], 'center': (cx, cy)}, ...], ...}

Licence

MIT — signa-core is the permissive, reusable core. The surrounding Signa WebODM plugin is AGPL-3.0 (WebODM-forced). This split mirrors the Itinera pattern (MIT core + copyleft plugin).

Note on the Signa worker

WebODM runs Signa's GCP detection by serializing a single self-contained function into a Celery worker, so the plugin keeps its own inlined copy of the detection primitive for that path. signa-core is the canonical, importable version for all non-worker consumers; the two are kept behaviourally identical.

Releasing (PyPI)

Published via PyPI Trusted Publishing (OIDC) — no API token is stored. See .github/workflows/publish-signa-core.yml and its header for the one-time pypi.org publisher setup. To release:

# bump version in pyproject.toml, then:
git tag signa-core-v0.1.0 && git push origin signa-core-v0.1.0

The workflow runs the tests, builds the sdist + wheel, and publishes them.