VLM output
An elevated CCTV camera overlooks a multi-lane riverside road at a signalized intersection with a crosswalk, where cars, a white coach bus, and numerous scooters move steadily through the frame. Traffic remains free-flowing and light-to-moderate across all four frames, with vehicles clearing the intersection and no stopped queues or incidents.
- Traffic Flow StateFree-Flowing
- Congestion LevelLow
- Queue DetectedNo
- Vehicle Count6-9 per frame
- Dominant Vehicle TypeScooters
- Intersection BlockedNo
- Signal VisibleYes (Red)
- Pedestrians On CrosswalkNo
- VisibilityClear/Overcast
Customer challenges
Congestion in view is noticed by manual monitoring, about 30 minutes late
Systems know traffic is jammed but cannot reason why (crash, roadworks, parking, signal fault)
Congestion cannot be predicted before it forms, so diversion is always reactive
Signal faults (dark, flashing, wrong display) in view are not detected; repair waits for reports
Vehicle entry, exit and travel time around parks and ports are not measured from existing feeds
Customer benefits
Congestion reporting time cut from 30 minutes to 3 minutes; congestion prediction enables pre-emptive diversion
VLM congestion analysis linked to dynamic signals improves peak travel time by 8% or more
Congestion insights pushed to the public 100+ times a month; signal-control strategy optimised 10+ times a month
Signal-fault repair time cut from 4–6 hours to 2 hours
Bus mobile inspection lifts network coverage 3.7x with zero new roadside equipment
Entry/exit data and diversion recommendations make peak traffic around key industrial corridors predictable
Key features
LPR — Entry/Exit Monitoring
Reads license plates to monitor vehicle entry and exit.
Traffic Analysis
Analyzes traffic volume and flow patterns.
Before & after
| Before | After — with Linker AI Nexa |
|---|---|
| Before Congestion known about 30 minutes late | After — with Linker AI Nexa Known in about 3 minutes, awareness time -90% |
| Before Fixed signal phasing | After — with Linker AI Nexa Dynamic timing linked to congestion nodes, peak travel time -8% |
| Before Public congestion information rarely pushed | After — with Linker AI Nexa 100+ pushes per month to CMS, navigation and radio (about 3 per day) |
| Before Signal-control strategy rarely revised | After — with Linker AI Nexa 10+ data-driven optimisations per month |
| Before Signal faults repaired 4–6 hours after report | After — with Linker AI Nexa Repaired in about 2 hours, -80% |
| Before 15% of intersections under monitoring | After — with Linker AI Nexa 55%+ with the bus fleet, coverage 3.7x |