Skip to content
Veritydocs
The science behind Verity

The method, weighed in the open.

Everything behind the calibrated likelihood ratio — the method, the open benchmark, the reference data, and the API. The app itself lives at verity.codes.

Start where you fit: scientists → the method · developers → use Verity · examiners → why it exists.

The method

How a scan becomes a calibrated likelihood ratio — leveling, signature extraction, alignment, congruent matching regions, and calibration — step by step on real marks.

Why Verity

What was missing: one calibrated, explainable, open method across mark types, where the statistics decide and the cost is characterized — not a black-box match.

The open benchmark

Frozen, source-disjoint splits scored on Cllr, calibration loss, and AUC. A leaderboard anyone can submit to, with a downloadable replication kit.

The data catalog

Browse the forensic scans behind the references — striated, impressed, and toolmark — content-addressed and downloadable as raw X3P.

References

The foundational methods, standards, benchmark datasets, and calibration literature Verity builds on, with relevance notes.

Scientific lineage

The published research Verity operationalizes — automatic bullet-land matching, Congruent Matching Cells, and the open-data ecosystem — plus the downloadable validation report.

Use Verity

Quickstart, core concepts (LR, Cllr, CMR), the REST API and reproducibility, the X3P codec, and the system architecture.

API reference (opens in new tab)

The interactive OpenAPI reference for the comparison API — endpoints, schemas, and live request examples.

One project, five hosts

Verity is one open codebase deployed across five subdomains — each does exactly one job.