CheckID
CustomersPartners
VIEN
Contact
PERFORMANCE SCIENCE

Biometric performance evaluation

Metrics are published as scientific/standards concepts, not Mobile-ID measured values unless a corresponding test report exists.

Talk to an expertTechnical documentation
◈
SOURCEAuthoritative / document
◈
CAPTUREDevice / SDK
◈
VERIFYDocument + biometric
◈
ASSUREPolicy + risk
◈
DECIDEClaims / status
◈
EVIDENCEAudit / trace
Technical basisSOURCE-BACKED
SourceCHECKID source register
StandardRoute-specific standard mapping
Version
ApplicabilitySee page scope
EvidenceQA evidence register
TECHNICAL REFERENCE

Biometric performance: a metric is meaningful only with test context

Operation → population → capture → threshold → metric → uncertainty → limitation

Verification

FMR/FNMR must be reported at a defined operating point or threshold. A standalone “accuracy” figure cannot compare systems when population, capture conditions and thresholds differ.

Identification

FPIR/FNIR/Rank-N depend on gallery/probe protocol, gallery size and search configuration. 1:N throughput must be tied to hardware/software configuration and workload.

PAD

APCER/BPCER are PAD metrics and are not interchangeable with matcher FMR/FNMR. PAD reports should identify tested attack-instrument classes/configuration when the source permits.

Benchmark evidence

A CheckID benchmark is publishable when it includes engine/model version, SDK, dataset/population, N, device, capture condition, threshold, hardware/software, methodology, test date/evaluator, result and limitations.

Current evidence status

The v8/v9 ZIP does not contain a benchmark report sufficient to publish numeric CheckID FMR/FNMR/FPIR/FNIR/APCER/BPCER or throughput. Current documentation keeps the metrics methodological and records the publication limitation.

Standards lifecycle

ISO/IEC 19795-1:2021 remains the published edition while ISO is developing Edition 3 at DIS stage. The site distinguishes current published references from draft revisions.

Authoritative technical references