A tool that generates PDF reports based on AI detections from your security cameras, integrated with the Frigate NVR system.
- Integration with Frigate NVR: Leverages local AI object detection capabilities to analyze footage and generate detailed reports [1].
- PDF Reports: Creates comprehensive PDF documents summarizing detected objects and actions.
- Automatic Timed and requested report generation.
- Built in pdf rendering (currently requires manual running.)
- Web page with report history and manual generation controls.
- Clone the project.
- Edit Settings.toml and insert the url for your frigate instance
- In your frigate config make sure to enable genai and put this in your prompt
Respond only in this json schema, do not include newlines and do not escape quotes. ( "type": "object", "properties": ( "threat_level": ( "type": "integer", "description": "user name" ), "suspiciousness": ( "type": "integer" ), "interest": ( "type": "integer" ), "description": ( "type": "string") ), "required": [ "threat_level", "suspiciousness", "interest", "description" ] )' - Make sure that Ollama is running and available on your local machine with the deepseek:32b model pulled (this can be changed in the source code but should be moved to the config).
- An active Frigate NVR setup with supported AI accelerators for object detection [1].
- Cameras configured with IP addresses or RTSP streams (as supported by Frigate)
- Compatible hardware to handle local processing of AI models.
- Local Ollama instance with deepseek:32b pulled.
