Smoke or fire
Early detection of smoke or open flame in the camera frame — before the situation turns into a fire.
Smoke or fire caught before it becomes a blaze
The "Smoke or fire" scenario detects smoke or open flame visible in the camera frame — catching the start of a fire at an early stage, often before point ceiling detectors react. In large halls, outdoors and in high-ceiling areas smoke can disperse far from a sensor, while image analysis covers the whole observed space.
Automatic detection lets you react the moment a hazard appears — instead of reconstructing the event from footage afterwards. Every second gained on an early response limits losses and risk to people.
Reaction at an early stage
A detection raises an alert and andon signal the instant smoke or flame appears in the watched zone.
Whole-space coverage
Analysis covers the full frame, including spots far from ceiling sensors — halls, high-ceiling areas and outdoor spaces.
Around-the-clock monitoring
Detection runs continuously, 24/7, with no fatigue or breaks — fire risk is covered on every shift.
How smoke-or-fire detection works
Industrial cameras watch the zone while vision servers run the SAM3 detection model with the prompt "smoke or fire" at the edge. Every confirmed match becomes an event that integration gateways forward to AndonCloud and machine PLCs.
The event turns into an alert, an andon signal, a dashboard entry and a line in a report — with no footage sent to the cloud and no strain on your infrastructure.
Detection at the edge
The model runs locally on the vision server — low latency and no footage streamed off-site.
Andon & PLC integration
A detected event can trigger a light signal, a duty-officer notification or an action on a machine controller.
No footage in the cloud
Analysis happens at the edge — only events leave the site, not the video stream.
Example use and privacy
In a paint shop a spark causes material to smoulder in a corner of the hall, far from the nearest ceiling sensor. The camera detects the smoke, AndonCloud immediately notifies the duty officer and points to the affected zone, giving time to intervene before the fire spreads.
The system recognises the "smoke or fire" phenomenon, not the identity of people — it performs no biometric identification or face recognition. The detection serves only to respond to fire risk.
No personal identification
Detection is about a phenomenon, not personal data — no face recognition.
Points to the affected zone
The alert shows which zone the hazard started in, cutting the time it takes to reach the source.
A full event trail
Every detection is logged with time and zone — ready for analysis in safety reports.
System
Our system is intuitive, but you can always count on us
We help your team work more efficiently every day.
Documentation
Customer support
Integrations
FAQ
Frequently asked questions
- The presence of smoke or open flame visible in the camera frame. The system recognises this phenomenon, not the identity of people.
- Image analysis covers the whole observed space, including spots far from ceiling sensors — in large halls, outdoors and under high ceilings it flags the hazard where it actually starts. It complements existing fire-protection systems rather than replacing them.
- No. Detection is about the smoke-or-fire phenomenon — there is no biometric identification or face recognition.
- The event triggers an alert, an andon signal, a dashboard entry and a report line; it can also trigger an action on a machine controller.
- Yes. The model runs at the edge on the vision server, so the reaction happens as the hazard appears.
- Standard industrial IP cameras watching the zone; the image is analysed locally by the vision server.
- No. Analysis happens at the edge — only events leave the site, not the video stream.
- Book a demo — we will pick the detection zones and show how the camera image becomes an alert and a decision in AndonCloud.
This scenario detects smoke or open flame visible in the camera frame — catching the start of a fire at an early stage, often before point smoke detectors react. In large halls, outdoors or in high-ceiling areas smoke can disperse far from a ceiling sensor; image analysis covers the whole observed space and flags the hazard where it actually starts.
Industrial cameras record continuously, and a vision server runs the SAM3 model with a configurable scenario at the edge (prompt: "smoke or fire"). Once a detection is confirmed, the integration gateway passes the event to AndonCloud, where it becomes an andon signal, an alert and an entry in dashboards and reports, and if needed can trigger a response on the machine PLCs. The system detects the phenomenon, not people's identity — it performs no biometric identification.
Example use: in a paint shop a spark causes material to smoulder in a corner of the hall, far from the nearest ceiling sensor. The camera detects the smoke, AndonCloud immediately notifies the duty officer and points to the affected zone, giving time to intervene before the fire spreads. The detection serves only to respond to fire risk.
Contact
Contact us
If you have any questions, talk to our expert
We will respond within the next business day
sales@andoncloud.comPhone
Monday - Friday; 9:00 - 17:00
+48 71 340 70 15
Marcin Wierzbicki
During the meeting, we will present the product and help you configure the system in your company efficiently and quickly. Our team is at your disposal.
Newsletter
Sign up for the newsletter
Want to stay up to date? Sign up for our database.
By subscribing to our newsletter, you agree to our Privacy Policy and to receive updates from our company.