CASE STUDY 13
Safety AI
INDUSTRIAL SAFETY
Safety risk management across existing CCTV infrastructure
A Korean steel manufacturer and other heavy industry sites · 2024 – present · operating infrastructure
CHALLENGE
- Large sites already carried extensive CCTV coverage, but the imagery was watched by people and therefore watched intermittently.
- Fire, smoke and safety incidents were detected by whoever happened to be looking, which is not a control.
- Replacing camera infrastructure to obtain analytics was not economically defensible at site scale.
SOLUTION
- Vivity Safety AI applied to existing CCTV, avoiding infrastructure replacement.
- Fire, smoke, and personal-safety detection running on-premise, with rapid hazard detection in defined work zones and compliance monitoring.
- Operators review AI-detected incidents in real time through the Safety AI Manager, across multiple cameras and recorders, with immediate alerting.
RESULTS
- Safety detection at an F-score above 0.98; fire and smoke detection above 0.95 in testing.
- More than ten simultaneous CCTV streams processed on-premise, with no imagery leaving the site.
- Designed to scale from a single work zone to estates of 1,000+ cameras.