Integration overview

Face Matcher exposes everything it produces through four surfaces: a REST API for commands and configuration, a GraphQL API for queries and real-time subscriptions, a RabbitMQ broker for message-based notifications, and an S3 store for images. Station, the bundled web UI, is itself a client of these surfaces, so anything Station can do, your own application can do as well. This section documents each surface from an integrator's point of view; how the recognition pipeline works is covered in the Face Matcher Server manual.

The four surfaces

Surface	From the host	Inside face-matcher-network	Typical use
REST API	http://localhost:8098 (Swagger UI at the root)	http://api:8080	Manage watchlists, enroll members, identify and verify faces, configure cameras and edge streams
GraphQL API	http://localhost:8097/graphql	http://graphql-api:8080/graphql	Query stored faces, tracklets and match results; subscribe to live events
RabbitMQ	localhost:5672 (AMQP), 1883 (MQTT), 5552 (streams), management UI on 15672	rmq:5672, rmq:1883, rmq:5552	Consume notifications from a message queue; receive edge-device frame data
S3 storage	http://localhost:8333	http://seaweedfs:8333	Read face crops, full frames and enrollment images by their image data ID

Station talks to the same endpoints: it reads CORE_API_ROOT=http://api:8080/api/v1 and GRAPHQL_ROOT=http://graphql-api:8080/graphql from .env.station, and it serves images to the browser through pre-signed S3 URLs. Authentication is off by default on all surfaces; the Authentication guide explains how to put the REST and GraphQL APIs behind OAuth2/OIDC.

Building a stack on top of Face Matcher

A product built on Face Matcher joins its Docker network and depends only on the items in the table below. Everything else in the deployment is internal and may change between releases. The Smart Corridor solution is built exactly this way: it vendors an unmodified copy of Face Matcher and relies on this integration list alone.

Item	Value
Network	face-matcher-network, created by run.sh. Join it with external: true in your own Compose file.
REST API	api:8080
GraphQL API	graphql-api:8080. Face templates are included in notifications (Notifications__IncludeTemplates=true).
RabbitMQ	rmq:5672 AMQP, rmq:1883 MQTT, rmq:5552 streams
S3	seaweedfs:8333. Use your own bucket; face-matcher belongs to the platform.
PostgreSQL	pgsql:5432
Station	fm-station:8000
Admin image	${REGISTRY}admin:${VERSION} from .env (database migrations and CLI tasks)
Credentials	All in .env (RabbitMQ section 2.2, S3 section 2.4, database section 2.1). The shipped values are development defaults; change them before production.
License	One iengine.lic in secrets/, see Get a license. Your stack does not need its own copy unless it runs a biometric engine itself.
Start order	Face Matcher first. start.sh reads STATION_IDENTIFICATION and STATION_PUBLIC_HOST from the environment.

A minimal Compose fragment for a service that joins the network:

services:
  my-service:
    image: example/my-service:1.0
    environment:
      - FACE_MATCHER_REST=http://api:8080
      - FACE_MATCHER_GRAPHQL=http://graphql-api:8080/graphql
      - RABBITMQ_HOST=rmq

networks:
  default:
    name: face-matcher-network
    external: true

Pages in this section

• REST API and the Enroll and identify via REST walkthrough.
• GraphQL API and GraphQL samples.
• RabbitMQ notifications and S3 storage.
• Enhanced preview for annotated video in third-party players and VMS.
• Auto-enrollment add-on.
• Compatibility matrix.

Face Matcher is delivered as a fixed-scope product: the APIs above are the extension points, and there is no custom feature work or one-off extension of the platform itself.