Pyronear
Spot wildfire starts early with cameras and a model that runs on site, close to the forest.
Pyronear detects fire starts early in natural areas. Its towers carry four or five high-resolution cameras and a micro-computer. Alerts reach a supervision platform built for firefighters. The project was born at Data for Good and its research is done by volunteers. It works with several French fire services, the French forestry office in Île-de-France and CENIA in Chile.
And where does AI come in?
The micro-computer captures one image per camera at regular intervals and analyses it on site, with a model trained to recognise smoke and flames. When it detects a fire, the images from that camera are sent to the project's database.
The AI label
- Model or technique
- Vision model trained on images of fire starts (Torch, Ultralytics, ncnn).
- Model published
- Yes
- Where it runs
- On site, on a Raspberry Pi-type micro-computer; images only leave after a detection.
- Data
- Images collected and annotated with partners and volunteers, published on Hugging Face under the Apache 2.0 licence.
- Role of people
- Not filled in
- Frugality
- A stated low-tech approach: a low-power device that can be adapted to cameras already in place.
“Not filled in”: MAUB could not check it with the project. Know more? Write to us.
Run this project? Show the “Listed on MAUB” badge
Paste this code on your site: an image and a link to this page. No script, no tracker: MAUB cannot tell who displays it.
<a href="https://maub.subalpine.it/en/projets/pyronear"><img src="https://maub.subalpine.it/statique/img/badge-recense-en.svg" alt="Listed on MAUB" width="180" height="48"></a>Something wrong or out of date? Write to contact@subalpine.it.
The monthly newsletter
Once a month, what moved on MAUB, in your inbox.