Live demo · road wear
Potholes, cracks and worn paint
Road condition is the layer councils ask about first, so here it is running rather than described. The clips below are unedited output from the same Vilnius drive: the detector marking defects and grading paint as the vehicle moves at traffic speed. Below them is everything it found across the whole route.
Clips are cut from the full annotated route videos, which run 5½ minutes per camera and ship at 1080p with the deliverables.
Everything found on one pass
Each finding carries a coordinate, so this is a work list rather than a folder of photographs. Toggle a layer to isolate it; click anything to see what was measured.
Defect volumes come from the LiDAR: the pipeline fits a plane to the surrounding road, measures the void beneath it, and reports the asphalt needed and the repair cut area. 715 of the 740 defects carry that measurement.
How the grading works
Two cameras, one finding
Front and rear cameras see the same stretch of road seconds apart. A defect confirmed by both is flagged as such, 227 of the 740 here, and that is what separates a real pothole from a shadow or a wet patch.
Measured, not just spotted
A detection alone gives you a dot. Fusing it with the point cloud gives depth, footprint area and void volume, so the output can be costed: 1.59 m³ of voids and 51 m² of repair cut across this route.
Paint graded, not just found
Markings are scored on coverage, contrast and retroreflectivity, then bucketed into good, worn and needs-repaint. That turns into 96 m of paint to redo now and 296 m worn or worse to schedule.
A sortable work list
Everything lands as GeoJSON and CSV in LKS-94 with LAS07 heights, sortable by severity or by street, so it drops into a maintenance plan rather than needing to be transcribed from photos.
Road condition surveys in Lithuania
A condition survey is only useful if it can be costed. Photographs of potholes tell a road maintenance team that a problem exists; a list with coordinates, depths, footprint areas and void volumes tells them what it will take to fix, and lets them sort the whole network by severity before a budget meeting.
AKYS surveys road condition across Lithuania as one layer of a mobile mapping pass, so it costs no extra driving. The same drive that produces the pavement assessment also produces the LiDAR point cloud the defect volumes are measured against, the 360° imagery that lets anyone audit a finding by eye, and the asset inventory covering signs, poles and street furniture.
Output is GeoJSON and CSV in LKS-94 with LAS07 heights, sortable by severity or by street, plus the annotated route video for verification. Lane markings are graded on paint coverage, contrast and retroreflectivity, which turns a repaint programme into a prioritised list of metres rather than a subjective walk-round.