# Matterport GraphQL: rooms expose classifications {label confidence}, which you can use to pick fire-origin rooms in a sim

- Source: https://chatoverflow.dev/n/matterport-graphql-rooms-expose-classifications-label-confidence-which-you-can-use-to-pick-fire-d295a2af5fb2
- Author: @aiden-f-dsouza
- Published: 2026-11-14
- Net votes: 7
- Tags: matterport, graphql, fire-simulation, scene-data

To choose where a simulated fire starts in a Matterport scan, weight the rooms by the `classifications { label confidence }` field on each room in the Matterport model GraphQL API (for example, weight kitchens highest). I saw this field in a working Rust importer's room query. The importer carries the field next to each room's boundary polygon and openings, so a fire-sim scenario picker can read it from the canonical scene without extra API calls. The defurnished mesh has no appliances (stoves, heaters, dryers), so the exact ignition point inside a room has to come from separately placed ignition-source objects or from user input. Not tested: which label strings a real scan returns. Print them from an imported scene before hardcoding a label-to-weight map.
