Talk to the team

Recognition

All on one camera.

Recognition reads the video from the cameras your plant already has and returns a verdict for each event, with evidence.

Talk to the teamSee how it works

One camera

Five readings of the same feed.

A single camera can run more than one scenario at once.

  • Fire and smoke detection

    Smoke and flame flagged in the video, with a clip of the moment.

  • PPE by body region

    Every region gets a verdict: compliant, non-compliant or undetermined.

  • Ergonomics

    Risky postures and time spent in each one.

  • Hazard zones

    Entry into marked areas and time spent inside.

  • Counting

    Crossings over a line or through an area.

PPE with three states

If the camera can't see a body region, the verdict is undetermined. No guessing.

  • compliant

    PPE is present on the checked region.

  • non-compliant

    PPE is missing on the checked region.

  • undetermined

    The region isn't visible enough to decide.

PPE by body region

Evidence, not assumptions

Every alert carries a clip of the moment and the marked frame. Whoever gets it sees what the system saw.

Fire and smoke

How it works

From camera to alert, inside the plant.

  • Your existing camera

    Works with the IP cameras already on site.

  • Edge device in the plant

    An NVIDIA Jetson runs the analysis on site. It keeps working offline; continuous video stays out of the cloud by default.

  • Verdict, evidence, alert

    Every event produces a verdict, an evidence clip and an alert for whoever needs to act.