PHYSICAL TEST ISOLATION · 11 BLIND-TEST GATES · FIXED COMMERCIAL THRESHOLDS

Zero test-set overlap.
Verified across real edge cases.

Before training SUNTEK, tens of thousands of blind-test sample hashes across 13 evaluation protocols were locked read-only and physically isolated from the training corpus, purging over 500,000 overlapping identities. Every metric below is measured under fixed commercial production thresholds and Out-of-Distribution (OOD) protocols.

HASH-ISOLATED BLIND SET
90K+ISOLATED

Physically isolated blind-test samples

Over 500,000 overlapping identities purged via strict hash deduplication across tens of millions of compliant training images.

50K OPEN-SET MASKED
99.97%

Upper-Zone Subspace TPIR @ FPIR=10⁻²

99.91% bare-face TPIR @ FPIR=10⁻³ and 99.97% masked TPIR on 50,000 enrolled + 20,000 unregistered impostors.

ISO 30107-3 PAD
0.00% APCER

Print, screen, Deepfake & toner

Measured at a single fixed commercial threshold (0.50) across visible-light nested PAD and 850nm NIR subsurface sensing.

DEMOGRAPHIC FAIRNESS
0.25% spread

99.80% mean across 4 groups

Balanced 24,000-pair multi-demographic blind audit (99.95% / 99.78% / 99.75% / 99.70% across 4 groups).

01COMPLETE G-01 ~ G-11 BLIND-TEST SCORECARD

Frozen Hold-Out Benchmarks + Horizontal Flip + Unseen OOD Audits

Eleven golden gates.
Every field challenge measured.

Instead of tuning separate thresholds per dataset, SUNTEK is evaluated under fixed commercial contracts across 50K open-set galleries, medical/N95 masks, heavy makeup, heavy bangs, disguise blocking, and dual-spectrum spoofing.

Gate IDEvaluation ScenarioConventional BaselineRelease Gate TargetSUNTEK Measured Result (with OOD & Flip)
G-01Cross-4-Demographic Fairness Blind Test (24,000 balanced verification pairs)Mean ~93% · Spread > 3.0%Mean ≥ 98.80% · Spread ≤ 0.60%Mean 99.80% · Max Spread 0.25% (99.95% / 99.78% / 99.75% / 99.70%)
G-0250,000-Person Open-Set 1:N + 20,000 Unregistered Impostor Stress TestOpen-set recall drops sharply under masksBare ≥ 93.8% · Mask ≥ 83.0%Bare TPIR@10⁻³ = 99.91% · Masked Upper-Zone Subspace TPIR@10⁻² = 99.97%
G-03Real-World Street Cross-Mask 1:1 Verification (Unmasked vs. Masked)Severe cross-mask drift at low FARStreet Cross-Mask TAR@10⁻⁴ ≥ 94.5% · Masked ≥ 97.0%Street Cross-Mask TAR@10⁻⁴ ≥ 99.9% (Acc 99.98%) · Masked Acc = 99.88% (TAR@10⁻⁴ ≥ 99.9%)
G-04Heavy Makeup ↔ Bare Face Cross-Modal Verification & Makeup LivenessElevated liveness false rejects under heavy makeupTAR@10⁻⁴ ≥ 98.0% · BPCER ≤ 0.40%TAR@10⁻⁴ ≥ 99.9% (Acc 99.95%) · Fixed-Threshold (0.50) Heavy Makeup Liveness BPCER = 0.00%
G-05Heavy Bangs Covering Eyebrows & Hairstyle Change Blind TestNotable cosine drift under forehead occlusionΔcos < 0.0150 · BPCER ≤ 0.40%Mid-Lower Subspace Δcos = 0.0017 (Full = 0.0032, Flip = 0.0078) · BPCER = 0.00%
G-06Access Wearing Compliance & Disguise Gate (Compliant Mask vs. Sunglasses, Half-Mask & Full-Veil/Half-Face)Disguise leaks & half-mask false alarmsMask recall ≥ 99.2% · Near-zero disguise passCompliant Mask Recall 99.33% · Sunglasses Block 98.93% · Dual/Full-Veil/Half-Mask/Half-Face Pass = 0.00% (@0.50)
G-07Full Beard & Multi-Skin-Tone Bare Face False-Mask ImmunityBeards & dark skin tones trigger false masksBeard ≤ 0.15% · Bare ≤ 0.15%Full Beard False Mask = 0.00% · Dark-Skin Bare Face = 0.09% (Unseen OOD Set = 0.00%)
G-08Extreme Frontal-Profile Pose & 30-Year Cross-Age Blind BenchmarksAccuracy degrades on steep yaw & multi-decade agingExtreme Profile ≥ 99.30% · 30-Year Age ≥ 98.30%Extreme Profile = 99.97% (Single 99.99%) · 30-Year Age = 99.98% · Standard 1:1 = 99.82% · Cross-Age = 95.78%
G-09Multi-Scale Face Detection, 98-Pt Dense Landmarks & Dynamic Eye BlinkBackground ghost boxes & coarse landmark errorAP ≥ 93%, 0 ghost boxes, NME -30%Multi-Scale Medium AP = 94.06% · Background Ghost Boxes = 0 · 98-Pt NME = 3.23% (-65.5%) · Blink F1 = 98.37% (AUC = 99.66%)
G-10Visible-Light Cross-CCT (2200K/6500K) Anti-Spoofing & Screen-Replayed Deepfake TestColor-temp drift & high-res Deepfake screen leaksAPCER ≤ 0.40% · CCT drift < 0.30%Fixed Threshold 0.50: 6500K/2200K BPCER = 0.00% · Physical & Deepfake APCER = 0.00% (Blue-Light Glasses BPCER 0.30%)
G-11Bimodal 850nm NIR Material PAD (OLED/LCD Screens, Laser Toner Print & 3D Silicone Masks)Screens, toner prints & 3D heads bypass RGB-only PADScreen 0.0%, Toner ≤ 0.2%, Silicone ≤ 0.6%850nm Real BPCER = 0.00% · OLED/LCD APCER = 0.00% · Laser Toner APCER = 0.00% · 3D Silicone APCER ≤ 0.55%

Note: All metrics above are measured on SUNTEK standard isolated blind-test sets under fixed commercial thresholds and do not constitute a guarantee for every environment or deployment. Real-world performance depends on ambient lighting, mounting geometry, and user cooperation—validate with field tuning on target hardware before production.

02ARCHITECTURAL PARADIGMS & OOD ROBUSTNESS

Design Fit Across Compute Budgets · SUNTEK Isolated Blind-Test Evidence

Built for each compute envelope,
tailored for edge access control.

Different face recognition architectures are engineered for different compute budgets and operating assumptions: heavy cloud networks maximize capacity on unconstrained server GPUs, lightweight global CNNs optimize general verification efficiency, and SUNTEK specializes in offline RK3576 NPU access control under partial occlusion.

Dimension / ScenarioLightweight CNNCloud Heavy (ResNet / ViT)SUNTEK (Edge 3-Zone)Design Trade-off & Fit
Edge NPU DeployabilityEngineered for low-power edge inference; lightweight quantized path suited to standard single-stream checksDesigned for floating-point server GPUs; large parameter footprint and global attention ops constrain embedded INT8 NPU deploymentEdge-quantized 101-layer backbone (24.18 GFLOPs; edge-cloud cosine fidelity ≥ 0.996)Server-class networks excel in centralized datacenters; SUNTEK natively fits RK3576 dual-core NPU offline scheduling
Standard Unoccluded 1:1 VerificationMature and stable commercial verification accuracy on standard frontal unoccluded facesDelivers top-tier representation capacity across standard unoccluded portrait benchmarks99.82% (Single-frame) / 99.78% (Multi-frame fused) on isolated blind testsMatches server-grade baseline precision on clean frontal captures within a compact edge compute budget
Large-Angle Profile & Cross-Age SpanConstrained feature capacity leads to gradual drift under steep yaw angles and multi-decade agingHigh upper bound in unconstrained wild poses, complex age spans, and high-capacity server matching99.97% Large-angle profile · 99.98% 30-yr age span · 61.40% Complex-pose TAR@10⁻⁴Bounded isometric geometry maintains turnstile capture stability and minimizes periodic gallery re-enrollment
50,000 Open-Set Masked SearchCoupled full-face vector representation without dedicated partial-occlusion subspace isolationTrained for holistic unoccluded semantics; strict-threshold masked search experiences lower-face dilution unless adapted99.97% Masked Upper-Zone (FPIR=10⁻²) / 99.91% Bare-Face (FPIR=10⁻³)Difference stems from task assumptions: cleanroom turnstiles require wearing-aware routing and orthogonal upper-zone isolation
Heavy Bangs Feature Preservation (Δcos)Global feature vectors shift when eyebrows and forehead are heavily coveredDeep global receptive fields or holistic attention propagate forehead changes across the full embedding0.0017 (Mid-Lower Subspace) / 0.0032 (Full 512-D)Tri-band orthogonal routing automatically relies on unoccluded mid-lower facial bands when heavy bangs appear—no re-enrollment needed
Access Gating & Dual-Spectrum PADTypically operates as a standalone feature extractor requiring separate external quality and liveness modulesFocuses on server-side static or single-stream comparison rather than terminal dual-ISP sync and voice compliance cues7 in-memory co-scheduled engines (ISO 29794-5 quality, half-mask coaching, RGB + 850nm NIR PAD)Real-world turnstile reliability depends equally on full-pipeline optical co-tuning, wearing governance, and liveness defense

Architecture Comparison Note & Objective Fairness Statement: This matrix compares engineering trade-offs across neural network paradigms from a system-selection perspective and does not target, map to, or evaluate any specific third-party vendor, commercial product, or open-source repository. Both lightweight global CNNs and heavy cloud networks perform strongly within their intended unoccluded or high-compute environments. All numerical results above reflect only SUNTEK self-measured benchmarks on internal physically isolated test sets under fixed commercial thresholds; field performance varies with lighting, mounting geometry, and hardware setup—validate on target devices prior to deployment.

OOD IMAGE DEGRADATION

Extreme JPEG Q=40, 5×5 Blur & Gamma 1.8

99.75% / 99.72% / 99.82%

Validated on unseen out-of-distribution test pairs degraded by heavy compression, optical defocus, and non-linear exposure shifts.

NO-REFERENCE QUALITY ASSESSMENT

International Quality Monotonic Ranking Correlation

+0.8805 SRCC

Pairwise monotonic ranking loss eliminates traditional MSE score inversion under severe blur, reaching +0.8805 SRCC against international quality yardsticks.

400–450NM BLUE-LIGHT DE-GLARE

Prescription Eyeglass Specular Immunity

1.40% False Alarm

Slashed anti-blue-light lens false sunglasses rejections from 21.96% to 1.40% while maintaining a consistent 98.93% true sunglasses block rate.

03GLOBAL REGULATORY COMPLIANCE & PRIVACY BY DESIGN

EU AI Act · GDPR · ISO/IEC 29794-5 (OFIQ) · ISO/IEC 30107-3 · PIPL

Engineered for enterprise audits
and global data sovereignty.

In multinational tenders and critical infrastructure deployments, algorithmic fairness, statutory quality standards, and zero-leakage privacy architecture are foundational.

01EU AI Act & ISO/IEC TR 24027

Algorithmic Fairness & Zero-Sensitive-Attribute Architecture

Designed from inception around data minimization and non-discrimination: the architecture inherently contains zero demographic or sensitive-attribute inference modules, extracting only anonymized identity geometry. Balanced 24,000-pair audits show 99.80% mean accuracy with only 0.25% spread across groups.

02ISO/IEC 29794-5 (OFIQ) & ISO/IEC 30107-3

Statutory Biometric Quality & PAD Testing Protocols

Natively embeds ISO/IEC 29794-5 OFIQ Unified Quality Score (UQS) alignment inside the 21-D assessment module, and validates dual-spectrum liveness under ISO/IEC 30107-3 APCER/BPCER protocols at fixed thresholds.

03GDPR Art. 9 & Art. 25 / PIPL

Irreversible Vectors & "Vector-Only, Zero Raw Portrait" Privacy

Transforms camera frames immediately into one-way irreversible 512-D hypersphere vectors in "template-only, zero raw image retention" mode, protected by on-device encrypted storage and hardware-isolated access controls.

04Air-Gapped Intranet & Data Sovereignty

100% Local Offline Execution with Zero Cloud Egress

All 7 neural networks execute 100% locally on the terminal NPU with zero external cloud calls or biometric data transmission—ready for air-gapped government, defense, finance, and industrial networks.

05End-to-End Physical Isolation & Governance

Hash-Deduplicated Training Isolation & Lifecycle Traceability

Enforces strict hash-based deduplication and physical isolation between training corpora and blind-test suites, paired with signed model pack verification and full version lifecycle governance.

SUNTEK FACE AI

Start from the hardware.
Ship with in-house algorithms.