AI descriptions
camera.ui can use a generative-AI model to describe what happens on camera. It writes a short summary of each moment in an event: a title, what happened, a notification-friendly line, and a threat level. It names people, plates, and detected movement.
Names and plates come from what face recognition and plate reading actually confirmed, not from what the model thinks it can read in the picture. A description will not put a plate on a car that was never read.[1]
Choosing a provider
In Settings → Recordings, open the GenAI tab and pick a provider:
- Ollama. Runs a model locally on your own hardware. Set the API Base URL (for example
http://localhost:11434). - OpenAI or Gemini. Use a cloud model. Set your API Key.
Then set the Model (such as llava, gpt-4o, or gemini-2.5-flash), an optional Description Language, and how many images per event to send. Request Timeout is how long camera.ui waits for the provider, 60 seconds by default; local Ollama models often need more. Test Connection checks it works. Enable it per camera with the camera's AI descriptions toggle.
Moments and Episodes
The provider you pick under GenAI feeds two features, each with its own tab:
- Moments. Turn on Enabled here to get AI descriptions of what happens in an event. Custom Instructions takes free-text guidance added to the built-in prompt, for example what to pay attention to or how to phrase things. An event's descriptions go out as a single summary notification a short while after the last activity.
- Episodes. Bundles related events across cameras into one story, with its own Custom Instructions. See Episodes.
Privacy
With Ollama, everything stays on your own hardware. With OpenAI or Gemini, event snapshots are sent to that provider to generate the description.[2]