Detection you can audit
Every defect keeps the frame it was found in, so a rating can be checked against the image that produced it rather than taken on trust.
NovaPave
Road Condition Assessment Using AI
From road imagery to a prioritized maintenance plan.
AI-powered road condition assessment that detects, analyzes, and maps road damage from imagery for faster and smarter infrastructure maintenance.
Road condition assessment — district network
Condition by section
- Good
- Fair
- Poor
Network condition
- 46% Good
- 32% Fair
- 22% Poor
| Section | Dominant damage | Rating |
|---|---|---|
| NH-53 · 12+000 – 14+500 | Potholes, alligator cracking | Poor |
| SH-9 · 4+200 – 6+000 | Edge break, ravelling | Poor |
| NH-53 · 18+600 – 21+000 | Transverse cracking | Fair |
- Network assessed
- 128 km
- Defects detected
- 1,240
- Priority sections
- 6
Road condition platform
Condition data that stands up to a maintenance argument
NovaPave reads the road from imagery you already capture, rates what it finds, and puts it where the maintenance decision gets made.
Network, not one stretch
Findings roll up from individual defects to section and network condition, which is the level maintenance budgets are actually set at.
Built for the maintenance decision
The output is a ranked list of what to fix first, with the evidence and the extent attached — not just a folder of tagged images.
Damage detection
What the imagery is read for
A survey run becomes a defect register: potholes, cracking and surface damage, each marked on the frame it was detected in.
- Road imagery
- AI detection
- Damage classification
- Severity rating
- Mapped findings
- Maintenance report
Potholes
Surface failures picked out of the run and located.
- Pothole identification from imagery and video frames
- Extent marked on the source frame
- Location carried through to the network map
- Evidence frame retained with each finding
Road cracks
Cracking captured by pattern, not just presence.
- Longitudinal and transverse cracking
- Interconnected (alligator) cracking
- Crack extent traced along the section
- Annotation geometry kept with the defect
Surface damage
The wider deterioration that sits around the obvious defects.
- Ravelling and surface wear
- Edge break along the carriageway
- Patching and previous repairs
- Rutting and deformation indicators
Condition severity
Rated defect by defect, reported section by section
Severity is what turns a list of defects into a maintenance position — which road is failing, how badly, and how fast.
Severity per defect
Each detected defect is rated on its own, so a single severe failure is not averaged away by the good road around it.
Condition per section
Defect ratings roll up into a condition band for each section, giving one comparable figure across the whole network.
Deterioration over time
Re-surveys are compared with earlier runs, so a section that is degrading quickly is visible before it fails.
Condition bands, defect classes and severity thresholds are configured from the standards your authority applies. We do not substitute our own rating scale, and we publish no blanket detection accuracy — performance is reported per model version against your own evaluation set.
Damage mapping
Every defect on the network, not in a folder
Mapped findings are what let a district-level conversation happen: where the damage clusters, which corridors are worst, what changed since the last run.
From frame to chainage to network
A defect is recorded against its section and chainage, so the same finding reads correctly on a frame, in a section summary and on the network map.
- Defect
- Chainage
- Section
- Network map
Located to the network
Every finding resolves to a point on the road, not a filename.
- Chainage-based location for each defect
- Section and road-level grouping
- Condition-coloured network map
- Filter by damage type or severity
Geospatial context
The assessment sits on the same map as your other data.
- Road network and boundary overlays
- KML/KMZ boundaries for jurisdictions
- Defect clusters visible across a district
- Exports for GIS and reporting tools
Maintenance planning
The output is a work list, in priority order
Condition assessment earns its place when it changes what gets fixed first — so NovaPave ends in priorities, extents and evidence.
Assess
Detected damage and severity ratings are compiled for every section surveyed.
Prioritize
Sections are ranked by condition and deterioration so the worst is addressed first.
Plan
Each priority section carries its dominant damage types and the extent involved.
Report
Findings, evidence and priorities are exported for the maintenance programme.
Maintenance requirements
Each section carries the damage driving its rating and the indicative extent to be treated, so the requirement is stated in maintenance terms.
Evidence attached
Priorities keep the annotated frames behind them, which is what makes a claim reviewable by an engineer or an auditor.
Reporting and export
Condition summaries, defect registers and priority lists export for circulation and for the systems your programme already runs on.
Ready to turn visual data into actionable intelligence?
Send us a road survey run — imagery or video — and we will show you the NovaPave condition assessment built from your own network.
Talk to NovaMetrics