Face SDK

Faceplugin Face SDK: on-premise face recognition, standalone face liveness, and combined Recognition + Liveness for mobile and server (ports 8083 / 8084).

Overview

Faceplugin Face SDK covers face recognition, standalone face liveness, and recognition + liveness in one server process. Images are processed on the device or on your server — not in Faceplugin’s cloud. Recognition is evaluated on NIST FRVT (Face Recognition Vendor Test).

What it can do: Capabilities. Then pick a mode:

• Recognition — recognition-only server API on 8083 (faceplugin/face-recognition)

• Liveness — standalone face PAD on 8084 / mobile (no matching)

• Recognition + Liveness — mobile Identify + 2D liveness, and server match + /api/liveness on 8083 (recommended for eKYC)

Features

• Face Detection

• Face Landmark Detection

• Face Template Extraction

• Face Template Matching

• Live 1:N Identify (mobile)

• Liveness Detection (passive 2D on mobile Identify)

• Pose Estimation

• On-premise Face Recognition API (port 8083)

Modes

Start with Capabilities, then open a mode.

Capabilities

Recognition, liveness, and recognition + liveness in one place.

Recognition

Recognition-only server (no /api/liveness).

Liveness

Standalone face anti-spoofing on mobile and port 8084.

Recognition + Liveness

Mobile + combined server — recommended for eKYC.

Typical call order on mobile: setActivation → init → detect / extract template → store templates in your database → similarity or VideoWorker (live 1:N). Identify default 0.67. Liveness default 0.5.

Typical call order on Linux / Windows: GET /api/machinecode → POST /api/activate → POST /api/detect / match / similarity. There is no server-side 1:N gallery (POST /api/identify does not exist).

FAQ

Can face recognition work completely offline? Yes. After you activate with a license key, matching does not need the internet. See Capabilities.

Does this SDK include liveness? Mobile Identify includes passive 2D liveness. For standalone face anti-spoofing, use the Face Liveness Detection SDK. On the server you can use combined Recognition + Liveness, which provides both services on port 8083.

Is there a Face Recognition API? Yes — Linux Docker and Windows HTTP on port 8083. See Linux, Windows, and combined Face Recognition + Liveness Linux / Windows.

Does the server SDK support 1:N identification? The server SDK does not provide a built-in 1:N gallery or POST /api/identify endpoint.

Instead, store face templates in your own database. When you need to identify a person, compare the probe template against your stored templates using /api/similarity and determine the best match in your application.

Use cases

• Access control — enroll face templates, then live 1:N Identify at a door or gate (mobile)

• Attendance — enroll staff once; match camera frames against your person database

• App login — 1:1 selfie match against a stored template (similarity / POST /api/match)

• KYC selfie match — after ID OCR, compare portrait crop to a live selfie (with Document Recognition and optional Face Liveness)

• Fraud checks — still-image detect / match / similarity on your server (port 8083)

Related documentation

• Capabilities · Recognition · Liveness · Recognition + Liveness · ID Document SDK

• Request a License · Try it · FAQ · Glossary

• Combining products (eKYC) · SDK comparison

• Status codes