Rural
Narrow streets lined with trees, often in shade. Scores matched across the whole range, with a correlation of 0.98.
Quality
Our pavement scores, checked by independent engineers.
The National Center for Asphalt Technology at Auburn University graded 2,250 of our street images by hand, then compared its pavement scores with ours.
Cyvl finds and grades every distress to ASTM D6433, the standard engineers use to score roads.
The results
Within two points of the experts on rural and city streets.
Pavement scores run from 0 to 100, and ASTM D6433 allows two surveys to differ by 5 points. This is the average difference between NCAT's engineers and Cyvl in each place.
Urban
Parked cars, patches, utility covers, and heavy traffic. The closest match of the three, with a correlation of 0.99.
Suburban
Streets with a lot of crack sealant. NCAT's report traces much of the gap to how sealed cracks were counted: ASTM D6433 treats a sealed crack as one low-severity distress, not two. Correlation 0.90.
“Evaluation of automated PCI results found consistently strong agreement with our expert annotations, with differences typically within industry‑recognized tolerances.”
How the study worked
Same images, two sets of scores.
-
We sent the images
2,250 street images from a rural town, a dense city, and a suburb, each with its own mix of conditions.
-
Engineers marked every distress
NCAT's team traced each crack, patch, and pothole by hand and graded how severe it was, following ASTM D6433.
-
The scores were compared
Each stretch of road was scored both ways, then checked with three standard tests: Pearson correlation, Lin's concordance, and Bland–Altman analysis.