GitHub project Readme

Smart Traffic Intersection Agent#

The Smart Traffic Intersection Agent is a comprehensive traffic analysis service that provides real-time intersection monitoring, directional traffic density analysis, and Vision Language Model (VLM) powered traffic insights. It processes MQTT traffic data, manages camera images, and delivers intelligent traffic analysis through RESTful APIs.

The agent supports sliding-window analysis, sustained traffic detection, and intelligent management of camera images to enhance traffic insights.

The following figure shows the high-level architecture of the Smart Traffic Intersection Agent, showcasing its core components and their interactions with external systems.

ITT architecture

The Smart Traffic UI below shows how traffic and weather data is analyzed, and how summary and alerts are shown to the user.

traffic intersection agent UI

Components#

The Smart Traffic Intersection stack includes the following containerized services:

  • MQTT Broker (Eclipse Mosquitto message broker) - Message broker for traffic data

  • DL Streamer Pipeline Server - Video analytics and AI inference

  • Intel® SceneScape Database - Configuration and metadata storage

  • Intel® SceneScape Web Server - Management interface

  • Intel® SceneScape Controller - Orchestration service

  • VLM OpenVINO Serving - Vision Language Model inference

  • Traffic Intelligence - Real-time traffic analysis with dual interface (API and UI)

Key Integration Points#

  • MQTT Communication: All services communicate via the shared MQTT broker

  • Docker Network: Services discover each other via Docker service names

  • Shared Secrets: TLS certificates and auth files mounted from src/secrets/

  • Persistent Storage: Traffic data stored in Docker volume traffic-intelligence-data

  • Health Monitoring: All services include health check endpoints

Learn More#