A self-hosted smart camera hub for event-driven monitoring and analytics. Connect ONVIF/DVRIP IP cameras, detect motion, classify people via OpenVINO Model Server (OVMS), and view live streams and events in a React UI—fully containerized with Docker Compose.

• Camera control & onboarding: ONVIF PTZ and imaging controls; Wi-Fi pairing & onboarding via DVRIP (no vendor apps required).
• Event pipeline: GoCV motion detection + FFmpeg frame extraction → OVMS classification → MinIO object promotion (staging → detections/false-positives) → PostgreSQL metadata → positive detections published to clients via Server-Sent Events (SSE).
• Inter-process messaging: RabbitMQ for decoupled communication between analyzer, motion, and API services.
• Live viewing: RTSP/WebRTC streaming via MediaMTX; React UI with real-time event feed and PTZ controls.
• Deployment: Single docker compose up stack (PostgreSQL, Redis, RabbitMQ, MediaMTX, MinIO, OVMS, Prometheus, Grafana).

Cameras (ONVIF/DVRIP)
   └─> motion (Go + GoCV) ──> analyzer (Go + FFmpeg + OVMS)
           │                           │
           └───────── RabbitMQ ────────┘
                                  │
                          PostgreSQL (metadata) + MinIO (objects)
                                  │
                   api (Go + Chi + SSE)  ──>  ui (React)  ──>  User

1. Clone

git clone https://github.com/tomerab1/cam-hub && cd cam-hub

2. Environment

cp .env.example .env
# Fill in Postgres, MinIO, RabbitMQ, and OVMS settings

3. Run

docker compose up --build

4. Access

• API: http://localhost:5555
• UI: http://localhost:5173 (or as configured)
• MediaMTX streams: as defined in compose configuration

• api/ — REST + SSE endpoints for camera control and event streaming (Chi, Go).
• analyzer/ — Handles FFmpeg frame extraction, OVMS inference, MinIO promotion, and RabbitMQ messaging.
• motion/ — GoCV motion detection and frame triggering logic.
• infra/ — Docker Compose setup, MediaMTX, OVMS configs.
• ui/ — React app for live viewing, event log, PTZ controls.

• Prometheus exporters and Grafana dashboards (metrics and event visualization).
• Advanced UI filters, camera grouping, and event replay.
• Automated tests and profiling tools for multi-camera setups.

• Tested mainly with specific ONVIF/DVRIP camera models.
• OVMS configuration assumes person-detection-retail-0013 model.

MIT