Engine operational · ArcFace 512D · MiniFASNet · 3DDFA_V2

Biometric engine
for developers

Drop production-grade face matching, liveness detection and anti-spoof into any system via REST API. No ML infrastructure required.

REST API ArcFace 512D Anti-spoof CNN 99.96% 3D morphology LGPD / GDPR compliant On-premise available
Get API key — free View endpoints
512
ArcFace dimensions
99.96%
Anti-spoof accuracy
38k+
Mesh landmarks
5
Biometric models
<3s
Avg response
Use cases

Built for systems that already have cameras

BioSignature is a biometric engine — not a KYC product. Integrate it into your existing infrastructure.

🏥

Clinical & hospital access

Verify patient or staff identity at check-in. Anti-spoof prevents badge sharing. Integrates with existing door/terminal hardware.

match + antispoof
🏋️

Sports & anti-doping

Confirm athlete identity before sample collection. Combines face match + liveness to prevent substitution fraud. SPMD-validated.

match + liveness
🎓

Remote exam proctoring

Verify student identity before OSCE or online exam. Compare enrollment photo vs live capture. Log with audit trail.

match + 3D morphology
🔐

Physical access control

Replace PIN/badge with face match. Anti-spoof CNN blocks photo attacks. Threshold configurable per security level.

match + antispoof
⚖️

Legal & receivership

Verify claimant identity before document review or disbursement. Audit log with ref, score and timestamp for compliance.

match + audit trail
🏦

KYC pipeline component

Already have document OCR? Drop in BioSignature as the face match layer. Outputs cosine score + decision for your logic.

engine only · white-label
API Endpoints

Five endpoints. One integration.

Each endpoint is independent. Use only what your system needs.

POST
/biosignature
Full pipeline — anti-spoof + ArcFace + 3D morphology fused into a single BioSignature score and decision.
Returns: decision · biosig_score · scores · evidence · ref
POST
/match
ArcFace 512D face matching only. Returns cosine similarity score between two face images.
Returns: arcface_score · bfm_score · same_person · threshold
POST
/antispoof
MiniFASNet CNN liveness check. Classifies image as REAL or SPOOF with confidence score.
Returns: verdict · real_score · spoof_score · confidence
POST
/analyze
Full face analysis — landmarks, pose, age estimation, quality score. No matching required.
Returns: landmarks_106 · pose · quality · embedding
POST
/analyze3d
3D reconstruction — 38,365 dense vertices, BFM shape coefficients (40D), DECA shape/detail vectors.
Returns: bfm_40d · shape_100d · detail_128d · mesh
GET
/health
Engine status and loaded models. Use for uptime monitoring and integration health checks.
Returns: status · models · version · endpoints
Engine architecture

Five models. One score.

The /biosignature endpoint runs all five models in parallel and returns a weighted fusion score.

01

Anti-spoof

MiniFASNet CNN detects printed photos, screen replays and masks. Trained on 70k+ samples. If SPOOF → immediate REJECTED.

MiniFASNet · 99.96%
02

Neural identity

ArcFace ResNet50 generates a 512D embedding. Cosine similarity against reference. Threshold: 0.28 same person.

ArcFace · 512D · weight 40%
03

3D morphology

3DDFA_V2 reconstructs 38,365 dense vertices. BFM 40D shape coefficients compared across images.

3DDFA · 38k pts · weight 10%
04

Cranial geometry

DECA FLAME model extracts 100D shape parameters — bone structure unique to each individual.

DECA · FLAME 100D · weight 25%
05

Skin texture

DECA detail encoder captures 128D micro-texture — wrinkles, pores, grain. Highly discriminative.

DECA detail · 128D · weight 20%

BioSignature Score

Weighted fusion. APPROVED ≥ 0.72 · REVIEW ≥ 0.50 · REJECTED < 0.50. Override: any SPOOF → REJECTED.

Configurable thresholds
Code

Integrate in minutes.

Standard REST. JSON in, JSON out. Works with any language or framework.

// POST https://biokyc.tech4life.pro/api/match
// Header: X-API-Key: bky_your_key

{
  "image1": "data:image/jpeg;base64,...",
  "image2": "data:image/jpeg;base64,...",
  "consent": true
}
Pricing

Simple, transparent pricing

All plans include all 5 endpoints. Pay per call or go unlimited.

Starter
$39
/ month · 100 calls
  • All 5 endpoints
  • JSON responses
  • Anti-spoof + face match
  • 10 req/min rate limit
  • Email support
Get started
Enterprise
$199
/ month · 2,000 calls
  • Everything in Professional
  • Custom thresholds
  • On-premise deployment
  • SLA guarantee
  • White-label available
  • LGPD/GDPR audit docs
Get started
Pay-as-you-go — no subscription required
$25
10 calls
$2.50 / call
$80
50 calls
$1.60 / call
$220
200 calls
$1.10 / call

Credits never expire · Stripe · PIX · OKX Crypto · MEXC Crypto

Need full KYC with document + face + AML?

BioSignature powers the Identify Platform — document intelligence + biometrics + sanctions in one unified API.

Explore Identify Platform →

Start building today.

Free API key. 3 calls included. No credit card required.

Get free API key Talk to us

Frequently Asked Questions

What is BioSignature?

BioSignature is a biometric identity verification system that combines facial recognition, liveness detection, and document analysis to perform secure Know Your Customer (KYC) checks. It creates a unique biometric signature from multiple data points to verify that a person is who they claim to be.

How does face matching work with ArcFace?

ArcFace is a deep learning model that maps a face into a high-dimensional embedding vector. When a user submits a selfie alongside an ID document photo, ArcFace generates embeddings for both images and computes a cosine similarity score. A score above the configured threshold confirms the faces belong to the same person, providing highly accurate one-to-one verification.

What is liveness detection?

Liveness detection determines whether the biometric sample comes from a live person rather than a spoofing artifact such as a printed photo, screen replay, or 3D mask. BioKYC uses multi-frame analysis and challenge-response techniques to confirm genuine human presence before proceeding with identity verification.

What biometric data is collected?

BioKYC collects facial imagery from a live selfie capture and from the photo on the submitted identity document. These images are processed to extract facial embeddings (numerical vectors) used for comparison. No raw images are retained beyond the verification session unless explicitly configured by the integrating organization.

How accurate is the anti-spoof detection?

The anti-spoof module achieves high accuracy by analyzing texture, depth cues, and temporal signals across multiple frames. It reliably rejects common presentation attacks including printed photos, screen replays, and paper masks, while maintaining a low false-rejection rate for genuine users.

Is biometric data stored securely?

Yes. All biometric data is encrypted in transit and at rest. Facial embeddings are stored separately from personally identifiable information, and access is restricted through role-based controls. The system is designed to comply with data protection regulations and supports configurable data retention policies.