kosdk/mqtt-camera-ai-monitor repository overview

⁠MQTT Camera AI Monitor

MQTT Camera AI Monitor is a Docker-ready Node.js/TypeScript application that listens to MQTT channels for trigger commands, captures images from RTSP cameras using FFmpeg, analyzes them via OpenAI-compatible AI endpoints, and publishes results and images back to MQTT. It supports multiple cameras, automatic cleanup, graceful shutdown, and detailed logging, all configured via a YAML file.

⁠Features

• Monitors MQTT topics for trigger commands
• Captures still images from RTSP camera streams (FFmpeg)
• Analyzes images using OpenAI-compatible endpoints
• Publishes results and images as MQTT messages
• Supports multiple cameras and flexible YAML config
• Automatic file cleanup, status monitoring, and detailed logging

⁠Quick Start

1. Prepare your config file:
   Copy and edit config.yaml.sample from the GitHub repository⁠ with your MQTT, camera, and AI endpoint settings.

2. Run the container:

   docker run -d \
     -v /path/to/your/config.yaml:/usr/src/app/config.yaml \
     kosdk/mqtt-camera-ai-monitor:latest
   

   Copy

   (Optionally, set a custom config path with -e CONFIG_FILE=/path/in/container/config.yaml)

3. MQTT Usage Example:

   • Trigger analysis by publishing YES to <basetopic>/<camera>/trigger
   • Image and AI analysis responses are published on <basetopic>/<camera>/image and <basetopic>/<camera>/ai
   • Monitor <basetopic>/online for status

⁠Prerequisites

• MQTT broker
• RTSP cameras accessible from container
• OpenAI-compatible API endpoint

⁠Docker Compose Example

services:
  mqtt-camera-monitor:
    image: kosdk/mqtt-camera-ai-monitor:latest
    volumes:
      - ./config.yaml:/app/config.yaml
    environment:
      - CONFIG_FILE=/app/config.yaml
    restart: unless-stopped

⁠Logs & Troubleshooting

• Logs output to console by default.
• For common issues (MQTT, camera access, AI API, permissions), check config, connectivity, and container logs.
• Monitor <basetopic>/online for health/status.

⁠License

BSD 3-Clause License

For configuration samples and advanced usage, visit the project GitHub⁠.