About Firstplume.
Software that recognises the first smoke of a wildfire, in images from ordinary fixed cameras.
Vision.
01
Early.
A fire starts with smoke. Firstplume looks for the first plume on the horizon.
02
Reliable.
An alert has to mean something. False alarms are the benchmark, and we measure them strictly.
03
Human.
Firstplume signals. A person reviews every alert and decides.
What we build.
- Detection per image.
- A detection model assesses every camera image.
- The sequence.
- A second layer alerts only when something persists and grows or shifts.
- Verification.
- An extra step against false alarms from clouds, sun and mist. In development.
- Open landscape.
- Firstplume is being developed for heath, dune and mixed forest, including mist and low sun.
How we measure.
- Per camera, per day.
- False alarms are the measure, on real wildfires and on blind-checked smoke-free days.
- Fixed in advance.
- Every comparison is set before anything is measured, and is independently reviewed.
- Reported honestly.
- With uncertainty margins, without performance claims.
Stage.
A Dutch start-up in formation, in the research and development stage. There is no product on the market yet.
- now
Research and development
Detection model, second layer and evaluation set-up.
Pilots
On real sites, with land managers and emergency services.
Product
After the pilots.
Work with us.
Land managers, safety regions, fire services and research partners: for pilots, imagery and feedback.