mouna-presence-reasoner

Sensor fusion occupancy inference combines multiple local signals like Hue presence, lights, LAN context, and camera artifacts to generate a live snapshot of who is home and which room they occupy.

You can infer room-level occupancy automatically by fusing multi-sensor signals to produce structured fields such as inferred room, confidence, and reasoning without requiring manual room state input.

To refresh your smart twin state, run a quick refresh to update the apartment occupancy snapshot and read the resulting structured fields directly from the generated state.json.

Multi-sensor occupancy inference provides a confidence score alongside the estimated room and reasoning, reflecting the reliability of the fused signals regardless of the original signal source.

Occupancy inference uses local signals such as Hue presence, LAN context, and camera artifacts, allowing you to generate apartment occupancy estimates without relying on external cloud processing.

The apartment state refresh outputs a structured state.json file containing fields such as inference.room, inference.confidence, inference.reason, avatars, and energyGuard.