Calibrex 0.4.1 — LiDAR-camera evidence in five minutes

[rsasaki0109]

Calibrex 0.4.1 — know whether your LiDAR-camera calibration is trustworthy

Calibrex 0.4.1 turns the first LiDAR-camera experience into a no-download, ROS-independent evidence workflow. Install the wheel, run the strict demo, inspect the HTML report, and verify the digest-bound provenance bundle:

python -m pip install \
  "https://github.com/rsasaki0109/Calibrex/releases/download/v0.4.1/calibrex-0.4.1-py3-none-any.whl"

calibrex demo kitti-lidar-camera-evidence \
  --output-dir outputs/kitti-lidar-camera-evidence \
  --strict-assessment
calibrex validate outputs/kitti-lidar-camera-evidence/result.yaml
calibrex verify outputs/kitti-lidar-camera-evidence/bundle.json \
  --require-raw-recomputed

What the demo proves

• The wheel contains a deterministic, generator-bound KITTI-shaped Camera-LiDAR fixture; no ROS installation or dataset download is required.
• The reference projection detects 16 of 24 mandatory ±1 degree / ±0.10 m known-bad perturbation cases.
• All six falsification-policy gates pass without relaxed thresholds.
• The result, report artifacts, input files, and assessment are connected through schema-valid, SHA-256-bound provenance.
• The clean Windows wheel path completes in under one minute; the same wheel smoke passes in GitHub Actions on Python 3.10, 3.11, and 3.12.

The bundled fixture verifies Calibrex's evaluation and evidence contracts. It is intentionally synthetic and is not a physical-rig accuracy claim or a standalone LiDAR-camera calibration algorithm. Use --dataset-path with an officially downloaded KITTI raw sequence when evaluating real data.

Other changes since 0.4.0

• Camera-LiDAR SOTA audit and research pipelines with digest-frozen protocols.
• Evidence-first solid-state LiDAR workflows, public-data gates, and real calibration visualizations.
• Calibration CI, schema, adapter, and provenance improvements accumulated on main.

See CHANGELOG.md and the full comparison.

Release verification

• Release workflow and isolated wheel smoke: passed.
• Main CI: all Python 3.10/3.11/3.12 test, lint, type-check, schema, and wheel-smoke jobs passed.
• Documentation build and GitHub Pages deployment: passed.
• Wheel SHA-256: 29888b11596e748ea1f211112bca50cf66e9822593b47b8fa2dd2648297cb55e
• Source distribution SHA-256: 093760324b616560a041af675c418e863fb94eadade7b39d00944822ed79b498

---

This discussion was created from the release Calibrex 0.4.1 — LiDAR-camera evidence in five minutes.

[rsasaki0109]

If you try v0.4.1 on your own data, a short sanitized reply is enough:

• Sensor pair and dataset type
• Final assessment: PASS, WARN, or FAIL
• The most useful failing gate or known-bad probe
• Whether calibrex verify ... --require-raw-recomputed passed
• One result that matched or surprised your expectation

Please do not upload private sensor data or machine-specific paths. A redacted assessment summary is useful, including failures—failed falsification gates are product feedback, not demo noise.