Linux & Docker Installation

This guide describes step by step how to install and configure an NVIDIA Jetson Orin Nano for use with the AI-Detector (CowCatcherAI). All steps are fully detailed so that nothing is skipped.

1. Hardware Installation & Flashing the Jetson

2. Installation of Docker & NVIDIA Container Toolkit

Open a terminal on the Jetson and execute the following commands:

sudo apt update
sudo apt install -y docker.io
sudo systemctl enable --now docker
sudo apt install -y nvidia-container-runtime nvidia-container-toolkit
sudo nvidia-ctk runtime configure --runtime=docker
sudo systemctl restart docker

3. Download AI-Detector Container

Download the correct container for JetPack 6:

sudo docker pull ghcr.io/eschouten/ai-detector:main-jetpack6

4. Creating configuration files

Create the necessary files using the following commands:

touch compose.yml
touch config.json

File 1: compose.yml

Open the file with `nano compose.yml`. Paste the following content:

services:
  aidetector:
    image: "ghcr.io/eschouten/ai-detector:main-jetpack6"
    runtime: nvidia
    ipc: host
    ulimits:
      memlock: -1
      stack: 67108864
    volumes:
      - ./config.json:/app/config.json
      - ./sprong24.mp4:/app/sprong24.mp4
      - ./detections/:/app/detections
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]

File 2: config.json

Open the file with `nano config.json`. Paste the following content and enter your own RTSP details and Telegram token/chat ID. Add more terms if you like, this is the minimal config file:

{
  "detectors": [
    {
      "detection": {
        "source": [
          "rtsp://admin:YourPassword123@192.168.100.22:554/h264Preview_01_sub"
        ]
      },
      "yolo": {
        "model": "https://github.com/CowCatcherAI/CowCatcherAI/releases/download/model-V16/cowcatcherV15.pt",
        "confidence": 0.84,
        "frames_min": 4,
        "timeout": 6,
        "time_max": 50
      },
      "exporters": {
        "telegram": [
          {
            "token": "<your_bot_token>",
            "chat": "<your_chat_id>",
            "alert_every": 5,
            "confidence": 0.87
          }
        ],
        "disk": {
          "directory": "mounts"
        }
      }
    }
  ]
}

5. Starting and Managing the Container

Start the container:

sudo docker compose up

Stop the container:

sudo docker compose down