VLM output
A forward-facing camera on a bus travels along a dedicated bus lane on an urban arterial road in an Asian city, with the badly cracked and patched lane surface filling the foreground. Traffic in the adjacent lanes (cars and scooters) flows in the same direction while the bus lane ahead stays clear, and the lane's painted markings become increasingly worn as the vehicle advances.
- Road Surface Damage DetectedYes
- Damage TypeAlligator Cracking
- Damage SeveritySevere
- Damage ExtentExtensive
- Pothole PresentNo
- Lane Marking ConditionFaded
- Affected LaneBus Lane
- Road Surface WetnessDry
- Maintenance PriorityHigh
Customer challenges
Bus and patrol footage passes over potholes daily but is never analysed
Size, depth and traffic impact must be confirmed by a site visit before dispatch
Jurisdiction (road, utility, drainage, district) is judged by hand, causing mis-routing
Post-storm pothole surges cannot be predicted or assessed quickly
Records lack image, location and dimensions, so hotspot analysis is impossible
Customer benefits
From passive complaints to proactive detection; pothole proactive detection rate rises from 10% to 70%
45 buses inspecting daily replace twice-monthly manual patrols; coverage density up 3.7x
Every alert carries image, coordinates, size and responsible agency; crews go straight to site with fewer surveys and reworks
Automatic cross-agency assignment cuts notification-to-dispatch time by 85%
Rapid damage assessment after storms and typhoons accelerates city recovery
Existing buses and cameras reused for low deployment cost; extends to litter, flooding and signal-fault scenarios
Key features
Pothole Detection (Bus-Mounted Cameras)
Detects potholes from bus-mounted cameras.
Pothole Attribute Reasoning (Size, Count, Traffic Impact, Manhole & Gutter Proximity)
Reasons pothole attributes including size, count, traffic impact, and proximity to manholes and gutters.
Jurisdiction Assignment (GPS + Road GIS)
Assigns jurisdiction using GPS and road GIS data.
Automated Dispatch & Case Closure (LINE, Maintenance System)
Dispatches cases through LINE and the maintenance system and tracks them to closure.
Pothole Hotspot Analytics & Post-Storm Inspection Mode
Analyzes pothole hotspots and supports post-storm inspection mode.
Before & after
| Before | After — with Linker AI Nexa |
|---|---|
| Before Proactive detection rate about 10% | After — with Linker AI Nexa 70%, a 7x rise in proactive discovery |
| Before Inspection twice a month | After — with Linker AI Nexa Daily inspection, coverage density 3.7x |
| Before Report to dispatch takes days | After — with Linker AI Nexa Dispatch time -85%, dispatch information accuracy +90% |
| Before Repair starts after an on-site confirmation | After — with Linker AI Nexa Repair turnaround as fast as 0.9 hours, most closed the same day |
| Before Post-storm backlog unknown for days | After — with Linker AI Nexa Peak 86 cases per day absorbed with no added inspection staff |
| Before Report accuracy unmeasured | After — with Linker AI Nexa ≥ 90% verified accuracy |