AI-Powered Infrastructure Monitoring from Municipal Dashcam Fleets
Challenge:The county's public works department relied on manual inspection to track infrastructure deterioration — inspectors drove periodic routes and logged issues by hand, creating a maintenance backlog that was perpetually weeks behind reality.
Solution:We deployed a fleet of vehicle-mounted dashcams across the county over a focused collection campaign, capturing 75+ hours of street-level footage covering every major road, intersection, and residential block.
Result:The enriched dataset became the training foundation for the county's automated infrastructure monitoring system.
The county's public works department relied on manual inspection to track infrastructure deterioration — inspectors drove periodic routes and logged issues by hand, creating a maintenance backlog that was perpetually weeks behind reality. With over 200 lane-miles of roads, dozens of intersections, and thousands of signs to monitor, the traditional approach couldn't keep pace. The county wanted to mount dashcams on existing municipal vehicles and use computer vision to flag issues automatically from the video feed. But that required training data: tens of hours of real local footage with labeled infrastructure defects, captured under the specific conditions those vehicles encounter — wet roads, overcast skies, night patrols, residential streets alongside commercial corridors.
We deployed a fleet of vehicle-mounted dashcams across the county over a focused collection campaign, capturing 75+ hours of street-level footage covering every major road, intersection, and residential block. Routes were planned for geographic completeness — ensuring diverse conditions including residential neighborhoods, commercial corridors, school zones, and industrial areas under varying weather and lighting.
Annotation combined automated AI enrichment with expert human review. First, a vision model analyzed representative frames from each clip to generate structured infrastructure condition reports and bounding box detections across 10 defect categories. Human annotators then validated and corrected every detection — verifying bounding box accuracy, reclassifying edge cases (a patched crack vs. active deterioration), and adding defects the model missed. This hybrid approach delivered the precision of human judgment at the speed of automated processing.
The detection taxonomy was co-developed with the county's public works team to match their existing maintenance categorization system — ensuring that model outputs map directly to work order types. Each clip's final annotation combined machine-generated bounding boxes, human-verified labels, natural language scene descriptions, and structured metadata — creating training pairs rich enough for the county's model to learn from.
The hybrid pipeline processed the full 75+ hour corpus in days rather than the months that purely manual annotation would have required, while maintaining the accuracy standard that safety-critical infrastructure monitoring demands.
Bellwood dashcam footage on Explore
Public municipal dashcam clips from Bellwood, Illinois, on Claru Explore, shown muted.
- Every detection human-validatedAnnotators verified bounding-box accuracy for each machine detection.
- Edge-case reclassificationReviewers separate cases such as a patched crack from active deterioration.
- Missed-defect passAnnotators add the defects the model missed.
- Taxonomy matches work ordersTen defect categories co-developed with public works to map onto existing work-order types.
Training pairs for the county's detection model.
- Geo-tagged dashcam footage
- 75+ hours, 455 GB, covering 200+ lane-miles.
- Bounding boxes
- Machine-generated detections, human-verified, across 10 defect categories.
- Scene descriptions
- A natural-language description of each clip.
- Structured metadata
- Per-clip metadata alongside the verified labels.
The enriched dataset became the training foundation for the county's automated infrastructure monitoring system. Municipal vehicles now capture footage as part of their normal routes — garbage trucks, utility vans, and patrol cars all contribute to a continuous street-level survey without any additional cost or staffing. The model flags defects in near real-time, automatically generating maintenance tickets categorized by severity and location. In the first month of deployment, the system identified 3x more infrastructure defects than the previous manual inspection process — the majority being cracked road surfaces and faded lane markings that would have gone unnoticed until they became safety hazards. The county estimates the automated system covers every road segment at least twice per week, compared to the previous quarterly manual inspection cycle.
Every municipality faces the same problem: infrastructure deteriorates faster than manual inspection can track it. Sending inspectors on periodic routes creates a permanently stale picture of road conditions — by the time an issue is logged, it may have already caused an accident or become far more expensive to repair. By layering AI inference on dashcam footage from vehicles already driving every street, cities get a continuously updated infrastructure health map at near-zero marginal cost. The critical enabler is high-quality, locally-specific training data — the same streets, weather patterns, and deterioration types the model will encounter in production. This engagement demonstrated that a focused collection campaign, enriched with automated vision analysis, can produce a production-ready training dataset for municipal AI deployment in days, not months.
Representative record from the annotation pipeline.
{
"clip_id": "a7f3e291-8c4d-4b1a-9e72-3d5f8a6b2c10",
"s3_key": "bellwood-dashcam/20260305_074341_00014.MP4",
"capture_date": "2026-03-05",
"gps": {
"lat": 41.9223,
"lng": -87.8704,
"speed_mph": 31
},
"frame_timestamp": "07:43:58",
"detections": [
{
"label": "cracked_road_surface",
"confidence": 0.87,
"bbox": [
518,
480,
615,
1440
],
"severity": "moderate"
},
{
"label": "faded_lane_marking",
"confidence": 0.79,
"bbox": [
518,
921,
626,
998
],
"severity": "low"
},
{
"label": "faded_lane_marking",
"confidence": 0.82,
"bbox": [
518,
1344,
626,
1440
],
"severity": "low"
},
{
"label": "overgrown_vegetation",
"confidence": 0.65,
"bbox": [
350,
1500,
580,
1920
],
"severity": "low"
}
],
"scene_description": "Overcast residential street with moderate road surface deterioration. Two-lane road with faded center line markings. Utility poles along right side. Sparse winter vegetation.",
"enrichment_source": "gemini_frame",
"annotation_status": "human_verified",
"work_order_mapping": {
"department": "Public Works — Roads Division",
"priority": "routine",
"estimated_repair": "crack_seal_and_remark"
}
}Ready to build your next dataset?
Tell us about your project and we will scope a plan within 48 hours.