Live demo · real inventory
Every street asset, found automatically
One drive through Vilkpėdės seniūnija produced 585 panoramas. Our detectors found 73,118 objects in them, and LiDAR fusion collapsed those sightings into 678 distinct real-world assets with a 3D position, a height and a footprint. The whole fusion step took 58 seconds. Click any point to see what the pipeline recorded.
Positions are LKS-94 (EPSG:3346) with LAS07 orthometric heights, delivered as GeoJSON, CSV and annotated panoramas. Parked vehicles and moving road users are tracked separately and are not shown here.
From pixels to a municipal asset register
Three detectors
A 115-class inventory model for street furniture, a road-user model for vehicles and pedestrians, and an open-vocabulary model that catches things no one thought to label in advance.
Seen from several angles
An asset is only accepted once it has been seen from multiple positions along the drive. Poles need three sightings within a tight spread, which is what removes the one-frame false positives.
Ranged by LiDAR
The detection gives direction; the point cloud gives distance. Fusing them yields a real 3D position, plus a height and a footprint, rather than a dot smeared somewhere near the kerb.
Straight into GIS
GeoJSON and CSV in the national grid, alongside annotated panoramas so anyone can audit a detection by eye and see the frame it came from.
Automated asset inventories for Lithuanian cities
Most municipal asset registers are assembled by hand, go stale within a season, and record a location accurate to whichever pole someone stood next to. An inventory built from a mobile mapping drive is different in kind: every tree, street light, traffic sign and billboard gets a coordinate in the national grid, a height, a footprint and a photograph proving it was there on a given date.
Re-driving the same route produces a directly comparable inventory, so change between surveys becomes a query rather than a project. That is what makes the output usable for maintenance planning, permit enforcement, green-space management and lighting audits.
AKYS runs these surveys across Lithuania and the Baltics. The detection models and deliverable formats are described on the AI detection and infrastructure inventory page; the capture behind them is mobile mapping in Lithuania with dual mobile LiDAR. Related layers from the same pass include road condition and Lithuanian sign detection.