Consultancy

3D Point Cloud AI in Production

Abderrazzaq Kharroubi

Abderrazzaq Kharroubi

Surveyor engineer and geomatician. PhD in 3D point cloud processing and AI.

See all products from Abderrazzaq

What this is:

A six-week implementation programme that takes your team from raw LiDAR or photogrammetric point clouds to a working AI classification pipeline they can run themselves on production data. The programme focuses on the four classification verticals that cover most operational LiDAR work: urban (buildings, ground, vegetation, vehicles), forest (canopy, ground, individual trees, biomass), rail (track, ballast, catenary, vegetation encroachment), and wires (powerlines, towers, vegetation around corridors). You pick the verticals that match your data. The work combines weekly working sessions with me, hands-on building of an open-source Python pipeline (PDAL, Open3D, modern point-cloud deep-learning frameworks), and structured upskilling of your team so they own the result. By the end of the six weeks your team can take a freshly delivered LiDAR project, run it through a reproducible pipeline, produce classified output for the chosen verticals, and ship the deliverable to a client or downstream system. You leave the engagement with the pipeline, the trained models, the documentation, and the team capability to extend both as your work evolves.

Who this is for:

  1. Surveying and geomatics firms (10–100 people) wanting to add AI classification to their existing LiDAR services without hiring a dedicated machine-learning team.

  2. Infrastructure operators (rail, utilities, transmission) processing recurring LiDAR captures for asset management or vegetation monitoring.

  3. Forestry or environmental consultancies running canopy, biomass, or stand-structure analyses at scale.

  4. Municipalities and mapping agencies wanting to run automated classification on their public LiDAR archives.

  5. Any team that has bought TerraSolid, LP360, or equivalent commercial software but is paying licence fees for AI features they cannot operate confidently.

What you get:

  • Week 1 — Discovery and pipeline architecture.

  • Weeks 2–3 — Pipeline build with your team

  • Week 4 — Model training on your data

  • Week 5 — Production hardening


Also included:

- Up to 2 hours of consulting per week (4 hours total) - Async email support between sessions - 7 days of follow-up support after the program

Contact

1-on-1 expert session to help your surveying firm start automating point cloud processing with AI.