Skip to main content

Solution Architecture

The solution is organized as four layers. Face Matcher — a product of its own — produces the identification signals at the bottom; the corridor's processing layer groups them into identities and zone behaviour; the Smart Corridor & eGate Hub turns them into clearance decisions and exposes them; thin UI clients consume them on top. For corridor events only the Hub is visible to integrators; Face Matcher additionally keeps its own public APIs and Station UI for camera and watchlist management.

Smart Corridor & eGate layered architecture

The three corridor layers and Face Matcher deploy together as one Compose-based stack, with Face Matcher vendored as a byte-identical release copy; deployment topologies are covered in Deployment Layouts.

Face Matcher and the integration boundary​

Face Matcher processes the camera streams, detects faces, matches them against watchlists and publishes every identification on its GraphQL, RabbitMQ and S3 interfaces. The corridor consumes it strictly through Face Matcher's published integration list — the face-matcher-network network, the container names and ports, the credentials, the license and the start order — and never modifies it. That boundary is what allows the vendored copy to be swapped for a separately deployed Face Matcher in a future release. See Face Matcher in the corridor and, for the engine itself, Face Matcher Architecture.

Processing & adaptation​

This layer turns Face Matcher's signals into the Hub's unified event model: the adapters translate their sources' output, while CIGS and CBS add the processing that groups identities and derives zone behavior. The optional MCT overlay adds cross-camera tracking as a further source.

ComponentRoleConnection
VPP AdapterTranslates Face Matcher identification eventsHub component
CIGS — Identity Grouping ServiceGroups cross-camera face detections into corridor identitiesVia CIGS Adapter
CIGS AdapterTranslates CIGS identity eventsHub component
CBS — Behavior ServiceDerives zone behavioral events (entry, dwell, exit)Publishes directly
MCT — Multi Camera Tracking (optional)Tracks people across cameras from an embedded-camera feedVia CBS

Without MCT, CBS derives the zone signals from Face Matcher and CIGS data.

Smart Corridor & eGate Hub​

The Hub is the central integration layer and the only system integrators interact with for corridor events. It aggregates and normalizes events from the internal components, decides clearance, persists the finalized outgoing events, and exposes them to external consumers. Internally, components communicate over a RabbitMQ exchange named biometric_events.

ServiceRoleStatus
GraphQL APILive subscriptions and data queries for external consumersActive
DbWorker ServiceEvent persistence and data retentionActive
REST APIAlternative HTTP interface for integrationsPlanned

UI​

On top sits the corridor dashboard — the Officer Monitoring App and the Traveler Display — which connects exclusively to the Hub's GraphQL API. See the Operational Interface for details.