odd-pi

Object Detection over Discord using the Raspberry PI

See Documentation for more details

stateDiagram-v2
    [*] --> DiscordServer: "Take Pic"
    DiscordServer --> DiscordAPI: Bot Listens
    DiscordAPI --> HomeServer: Coroutine Begins
    HomeServer --> LocalNetwork: raspistill
    LocalNetwork --> Pi: raspistill
    state Pi {
        [*] --> Camera
    }
    Pi --> LocalNetwork: pic.png
    LocalNetwork --> HomeServer: pic.png
    HomeServer --> DiscordAPI: Coroutine Ends
    DiscordAPI --> DiscordServer: Image Posted
    DiscordServer --> [*]: User Notified

Quickstart

• Setup Raspberry Pi with Camera and allow SSH
• On a different machine:
  • git clone https://github.com/jvivian/odd-pi && cd odd-pi && pip install ./
  • Set required env variables in .env file
  • python scripts/launch_discord_bot.py run_discord_bot --help
• Run an active camera at given interval which looks for a specified object and posts it to Discord
• Type pi! (or given keyword) in any channel to get a picture taken
• Type yolo! (or keyword) to have a picture with predictions overlayed

Roadmap

• Add CLI to launch script
  • camera commands, etc,
• Add critical CCC: Cat Camera Coroutine
  • Make extensible to any class \(C\) object, time duration, etc
• Avoid camera conflicts by making file names unique (or just deleting)
• Add console-script on install