Automated classification
Pre-trained AI for LV/HV wires, towers, insulators, guy wires, vegetation and ground, straight from raw LAS/LAZ.
Utility · Powerline corridor LiDAR
From raw LiDAR to grid-ready intelligence, in minutes not weeks.
Flai automates the full powerline corridor workflow: classification, vectorisation, vegetation clearance, conductor geometry, defect reporting and change detection, all in one platform, proven at national scale across Europe and North America.
200,000+
linear km of powerline corridor processed by Flai customers each year
F1 0.99
HV wire classification accuracy on benchmark powerline data
≤ 10 min
average processing time per linear km, fully unsupervised
200+
customers across 50+ countries, from utilities and DSOs to mapping companies
Capabilities
What Flai does for powerline operators.
Network vectorisation
Conductors and supports detected and vectorised into a full 3D network model, ready for geometric and clearance analysis.
Vegetation & clearance
Mass-automated conductor-to-vegetation, conductor-to-ground and conductor-to-structure distances across entire corridors.
Conductor geometry
Conductor sag, pole inclination, span risk scoring and per-defect violation visualisation, all in one workspace.
Multi-source data fusion
Enrich LiDAR with aerial and satellite imagery (e.g. PlanetLabs) for tree-species classification, growth tracking and habitat context.
Adaptable to your assets
Extend the schema with custom classes: distribution transformers, switchgear, dampers or any object you care about.
On-prem retraining
Fine-tune the powerline AI on your own annotated data, on Flai cloud or fully on-prem. No data leaves your network.
Open formats & API
LAS / LAZ / COPC in, CSV / XLS / SHP / GeoTIFF out, plus a REST API and QGIS plugin for direct GIS integration.
Workflow
The corridor workflow, end to end.
Import LiDAR
LAS, LAZ, COPC, E57, PLY. All major airborne and UAV sensors, including DJI L2.
Pre-process
Noise and cluster filtering, OSM masks, ground refinement, flight-line cleanup.
AI classification
Pre-trained utility model: wires, towers, insulators, vegetation, ground.
3D network model
Conductor fitting, tower extraction, GIS-ID matching, span topology.
Reports & export
Clearance, sag, lean, defect catalog. CSV / XLS / SHP / API.
Classes the model resolves
Down to the conductor hardware.
Beyond the broad classes, the dedicated powerline model resolves the fine hardware that inspection and vegetation-management workflows depend on.
All 17 available categories
01Other02Ground03Vegetation06Buildings07Low noise18High noise21Vehicles24Low-voltage wires25High-voltage wires26Railroad wires27Low-voltage towers28High-voltage towers29Railroad towers30Fences31Insulators32Guy wires34Tension wire
Highlighted classes are the powerline-specific hardware. Colours match the Flai powerline classifier.
Application
Vegetation management.
Pair the segmented conductors with vegetation to compute conductor-to-vegetation distance automatically, flagging growth inside the clearance envelope. The result is a prioritised, span-by-span map: which trees threaten the line, where, and how urgently to cut.
Application
Asset mapping.
Every tower, conductor and piece of hardware becomes a typed, geolocated asset. Flai vectorises the corridor into a 3D network model with span topology and GIS-ID matching, so the classified cloud drops straight into your GIS as a queryable asset inventory, ready for inspection planning and maintenance.
- Conductor sag & ground clearanceObserved geometry on every span, including catenary fit and minimum ground clearance.
- Pole inclination & leanAutomated angle measurement and side-by-side comparison across survey epochs.
- Defect reporting & exportColour-coded violations in the 3D view, exported as CSV / XLS / SHP / API into GIS or ticketing systems.
- Change detectionMulti-flight comparison and DSM differencing to find new defects between captures.
Benchmark accuracy
Measured, not marketed.
F1 scores on benchmark powerline data, straight from the pre-trained utility model.
| HV wires | F1 0.99 | |
| Ground | F1 0.98 | |
| Buildings | F1 0.97 | |
| Vegetation | F1 0.95 | |
| Towers / poles | F1 0.93 |
"Using Flai cut down the time for our manual point cloud classification by a factor of four, radically improving our turn-around and project profitability."
Deployment & integration
Run it wherever your data must live.
Flai-managed Cloud
AWS EU-Central (Frankfurt) by default. Production-grade, monitored, 99.9% uptime SLA. Onboarding in days.
Customer Cloud / Self-hosted
Dedicated deployment in your own GCP, AWS or Azure. Full data sovereignty, separate release pipeline.
On-premise / Air-gapped
The Flai CLI runs entirely behind your firewall. Air-gap supported, with only metadata for licence validation.
Classify powerline corridors faster.
From raw LiDAR to a grid-ready 3D network, at national scale. Cut manual work and deliver consistent, accurate results.