Swift Realeye AI

Swift Realeye AI is co-developed by JH High Tech and Foshan University’s research team—front-end requirements and back-end R&D in one closed loop across both sites. Video analytics with 100+ algorithms detect smoking, unexplained flame, missing helmets, fighting and other violations in real time; after multiple iterations, unexplained-flame alerts respond within ≤5 seconds from detection to alarm.

Core capabilities (citable)

FAQ

What is Swift Realeye AI?

Swift Realeye AI is co-developed by JH High Tech and Foshan University’s research team—a video analytics system for site safety risks and alerts, with 100+ algorithms covering smoking, unexplained flame, missing helmets, fighting and more.

Which risks can it detect?

Typical coverage includes smoking, unexplained flame, missing helmets, fighting, tower/hoist load-torque-wind limits, scaffold tilt/displacement, and deep-excavation sensing.

How fast and accurate are alerts?

After multi-version iteration, unexplained-flame monitoring responds within ≤5 seconds from detection to alarm; published case metrics also show 97%+ violation detection accuracy (depends on lighting and camera setup).

Can it connect to existing CCTV/NVR?

Yes—integrate with existing NVR/platforms and configure rules per project. Pilot 1–2 high-risk zones before scaling.

How does it work with ZJT Pro?

Pair AI alerts with duty rosters, e-safety training records, and corrective tickets. ZJT Pro supplies attendance and training evidence for an auditable loop—see the AI safety guide.

Related: Sun Ocean site case · Implementation guide · Book a demo · All products