Skip to main content

Getting started

This group takes you from an empty Linux host to your first face identification: check the hardware, obtain a license, install the package with Docker Compose, then register yourself in a watchlist and walk in front of a camera. The default deployment runs every service once (the camera service five times) on a single host with CPU processing, which is right for evaluation and for small production sites; scaling beyond that is covered in the Guides.

Prerequisites​

  • A Linux host that meets the Hardware requirements: an x86_64 or ARM CPU with AVX2 (x86_64), at least 4 physical cores, 16 GB RAM and 80 GB of storage.
  • Docker Engine with the Docker Compose plugin. Face Matcher is Docker Compose on Linux only; there is no Windows package and no Kubernetes deployment.
  • Credentials for the Customer Portal, provided by your sales representative. They give you the container registry login and let you generate a license.
  • Administrator-level familiarity with Linux and basic Docker knowledge (images, containers, Compose, registries).
  • For the first identification: at least one IP camera with an RTSP stream, or a smart camera running the Embedded Stream Processor, or just a webcam and a photo of yourself for API-only identification.

Steps​

  1. Hardware requirements: confirm the host is adequate for the number of cameras or requests you plan, and read Camera selection and placement before mounting cameras.
  2. Get a license: read the host's hardware ID with the license manager image and generate iengine.lic in the Customer Portal. The license is bound to that host.
  3. Install: log in to the registry, put the license in secrets/, run start.sh and open Station.
  4. The release package: understand what each file configures before you change anything, and how to keep your production copy upgradable.
  5. First identification: create a watchlist with your own photo, add a camera and see the match in Station.

Once that works, the Face Matcher Server manual explains what happened under the hood and the Integration pages show how to receive the same event in your own software.