Docs / Face SDK Docs
Face SDK Docs
Install the ARSA Face SDK on your own server, enrol people, verify and identify faces, check liveness and connect your systems through the API.
The ARSA Face SDK is face recognition and liveness detection that runs on your own server, as a Docker container with an HTTP API and a web dashboard. Photos, faces and names stay on your server.
What It Does
- Verify (1:1): compare two photos, or a photo against an enrolled person, and get a match decision with its score and threshold.
- Identify (1:N): find who is in a photo among the people you enrolled, for every face in the picture.
- Liveness: check that a face in front of the camera is a live person, and not a printed photo or a screen. Passive liveness works on ordinary photos; active liveness (Prod) adds a short challenge.
- Cameras and events (Prod): recognise people on RTSP cameras, watch them in the dashboard’s Live view and receive events by WebSocket or webhook.
- Face attributes: optional gender and expression estimates, off by default.
- Dashboard: enrol people, try every call in the Sandbox, manage API keys and read the request and audit logs.
Editions
| Prod | Demo | |
|---|---|---|
| Use | Production, licensed per server | Free 14-day evaluation |
| Recognition | Server model (default) and light model | Light model |
| Liveness | Passive and active | Passive |
| Cameras | RTSP and RTSPS cameras, live video in the dashboard | The browser’s webcam in Live view |
| Limits | As licensed | 100 people, 50 API calls an hour |
| Activation | Online, or offline with a request file | Online, checks in once a day |
Where To Start
- Quick Start: run the Demo with one command and make your first call.
- Install Prod: production install, GPU images and cameras.
- First Run: create the owner account and your first API key in the dashboard.
- Authentication and the API reference: connect your own system.
Measured accuracy is published on the benchmarks page.
Was this page helpful?
API v2. Updated 2026-11-12