VLM output
Two utility linemen in blue uniforms and yellow hard hats work at the top of a concrete power pole, handling conductors and a transformer connection against an overcast sky. One worker is climbing/standing on the pole with a harness and tool pouch while the second works from the basket of an aerial bucket lift; both remain in position across the frames.
- Workers Detected2
- Helmet ComplianceYes
- Fall Harness WornPartial
- Working At HeightYes
- Aerial Lift In UseYes
- Live Line ProximityHigh
- Gloves WornYes
- Fall Or Incident ObservedNo
- Overall Risk LevelHigh
Customer challenges
Safety monitoring is quota-based manual watching; most footage is opened only as after-the-fact evidence
PPE compliance - harness actually clipped, insulating gloves worn, auxiliary rope alternated on the pole - has to be judged frame by frame by a person
Confined-space rules such as a tripod above the manhole, the ventilation duct overlapping and running, and headcount in and out cannot be watched continuously
Live-line work needs the de-energisation test and safe distance verified, which legacy CV models do not read
Cameras are misaimed, blocked or capture no worker, so the footage does not even qualify as evidence
Contractor performance is anecdotal: violations are not counted, classified or attributed to a unit
Customer benefits
1. About 100% monitoring coverage replaces quota-based manual watching of a fraction of the feeds
2. More than 50 AI models across the six statutory scenarios of the sector occupational-safety provisions
3. Violations alerted in real time under L1-L5 rules to the responsible foreman and unit, with on-site buzzer, push or call
4. Valid-frame detection rejects blocked or misaimed footage at capture, so the record stands as evidence
5. Contractor management database turns violations into monthly statistics, comparison and contractor scoring
6. Auto-labeling and Auto-ML platforms let the utility add scenarios and retrain models without a vendor cycle
Key features
PPE & Fall-Protection Compliance (Helmet, Harness, Auxiliary Rope, Insulating Gloves/Boots/Face Shield)
Checks PPE and fall-protection compliance, including helmet, harness, auxiliary rope, and insulating gloves, boots, and face shield.
Statutory Scenario Monitoring (Work at Height above 2 m, Live-Line & Near-Live Work, Hot Work & No-Smoking, Confined Space)
Monitors statutory scenarios such as work at height above 2 m, live-line and near-live work, hot work and no-smoking areas, and confined space.
Worker Abnormal Condition, Violation Behaviour & Valid-Frame Detection
Detects abnormal worker conditions, violation behavior, and valid frames for review.
Before & after
| Before | After — with Linker AI Nexa |
|---|---|
| Before 1. Monitoring coverage, quota-based manual watching of a fraction of feeds | After — with Linker AI Nexa 1. About 100% of feeds under continuous AI watch |
| Before 2. Violation awareness, after an incident is reported | After — with Linker AI Nexa 2. Real-time alert within seconds of detection |
| Before 3. Share of the site camera fleet running AI, pilot only | After — with Linker AI Nexa 3. 100% of the mobile camera fleet, 4.4x the pilot scale |
| Before 4. Statutory scenarios checked automatically, 0 of 6 | After — with Linker AI Nexa 4. 6 of 6, covered by 50+ AI models |
| Before 5. Share of footage verified as valid evidence, unmeasured | After — with Linker AI Nexa 5. 100% checked at capture, invalid frames rejected |
| Before 6. Quantified contractor violations, 0 case | After — with Linker AI Nexa 6. Monthly violation counts, categories and per-contractor scoring |