VLM output
A static camera across a body of water views a treeline where several separate ground-level fires burn along the forest edge, with open flames visible beneath the canopy. Over the four frames the flames persist and grey-black smoke thickens, spreading upward and laterally to obscure much of the sky and background trees.
- Fire DetectedYes
- Smoke DetectedYes
- Active Flame Count3-4
- Fire SeverityHigh
- Smoke DensityHeavy
- Air Quality ImpactSevere
- Fire Spread TrendIncreasing
- VisibilityReduced
- People Or Vehicles PresentNo
Customer challenges
Smoke and flame in view are not recognised in time; by arrival nothing is left as evidence
Steam and pollutant plumes are hard to tell apart; stack emissions need continuous visual judgement
Black-smoke vehicles must be recognised in motion with plate capture
Early wildfire and farmland fire smoke is not detected in time
Smoke source and drift direction are not reasoned; no hotspot or seasonal records
Customer benefits
Inspection shifts from after-the-fact complaints to real-time discovery, with the full picture and evidence before arrival
Stacks and hotspots monitored 24 hours a day, so offenders can no longer choose their timing
Higher proactive detection of black-smoke vehicles, with plates locked for direct citation
Early wildfires and farmland fires found sooner, protecting the golden firefighting window
Evidence packages generated automatically so inspectors focus on enforcement rather than evidence gathering
Existing intersection CCTV and bus cameras reused, sharing fleet and compute with litter detection
Key features
Open Burning Detection
Detects open burning in monitored areas.
Fire & Smoke Detection (Early Wildfire)
Detects fire and smoke for early wildfire warning.
Factory Stack Emission Anomaly Detection
Detects abnormal emissions from factory stacks.
Black Smoke Vehicle Detection (with LPR)
Detects black smoke vehicles and captures their plates with LPR.
Evidence Package & Emission Hotspot Analytics
Compiles evidence packages and analyzes emission hotspots.
Before & after
| Before | After — with Linker AI Nexa |
|---|---|
| Before Black-smoke vehicles caught by tips | After — with Linker AI Nexa Proactive detection +60% |
| Before Complaints arrive after the fire is out | After — with Linker AI Nexa Detected 60 minutes earlier (environmental scenario reference) |
| Before Hotspots patrolled a few times a week | After — with Linker AI Nexa 24/7 monitoring, violations at monitored hotspots -80% (dumping reference) |
| Before No evidence available on arrival | After — with Linker AI Nexa 100% of detections carry an evidence package with image, time, GPS and plate |
| Before Stack plumes checked on inspection days only | After — with Linker AI Nexa Continuous plume monitoring with steam vs. pollutant classification (accuracy target) |
| Before No systematic pollution event records | After — with Linker AI Nexa 100% of events logged with time, location and scale for hotspot and seasonal analysis |