AI for 3D Point Clouds Classification
Abderrazzaq Kharroubi
PhD in Geomatics · LiDAR & AI Specialist
Build a production-grade AI pipeline for 3D point clouds.
A 4-week live cohort to build a production-grade AI pipeline for 3D LiDAR point clouds, end to end. You build the full workflow: data preparation, feature engineering, classical machine learning baseline, deep learning with RandLA-Net, post-processing rules, and ASPRS-compliant LAS export. By week 4 you have a pipeline that runs on your own data and produces files clients can use.
The anchor application is power line corridor classification, the same workflow I built during my PhD on Belgian railway corridors. It transfers directly to forestry, urban mapping, mobile mapping, and rail.
What you’ll learn
Build a production-grade AI pipeline that classifies LiDAR corridors end-to-end and delivers ASPRS LAS files that clients can use.
Prepare point clouds for ML/DL: spatial indexing, ground filtering, downsampling, tiling
Compute geometric and height features that separate corridor classes.
Train and evaluate Random Forest and RandLA-Net classifiers on real corridor data
Post-process predictions and compute vegetation-to-wire clearance distances.
Learn directly from Abderrazzaq
Abderrazzaq Kharroubi
PhD in geomatics, contributor to research projects on 3D point clouds.
Who this course is for
A surveying or geomatics engineer delivering classified LAS to utility, rail or telecom clients, and tired of doing it by hand.
A GIS or remote-sensing professional who wants to add AI-based classification to your service offering.
A researcher or PhD student working on point clouds and ready to move past tutorial notebooks to a real pipeline.
What's included
Live sessions
Learn directly from Abderrazzaq Kharroubi in a real-time, interactive format.
Certificate of completion
Share your new skills with your employer or on LinkedIn.
Cohort community
Ask questions between sessions, share your results, get feedback.
Free signed copy of my upcoming book
Founding-cohort participants get a complimentary signed copy of my upcoming book on AI for 3D point clouds, shipping when published.
Maven Guarantee
Your purchase is backed by the Maven Guarantee.
Course syllabus
7 live sessions • 9 lessons
Week 1
Mar
8
Live 1.1
Sun 3/811:00 PM—12:30 AM (UTC)Mar
10
Live 1.2
Tue 3/1010:30 AM—12:00 PM (UTC)Mar
11
Live 1.3
Wed 3/1111:00 AM—12:00 PM (UTC)Mar
13
Live 2.1
Fri 3/1310:30 AM—12:00 PM (UTC)
LiDAR technology and 3D point clouds
Point cloud processing fundamentals
Data preparation and annotation
Feature engineering
Week 2
Schedule
Live sessions
16-18 hrs
4 weeks
Sun, Mar 8
11:00 PM—12:30 AM (UTC)
Tue, Mar 10
10:30 AM—12:00 PM (UTC)
Wed, Mar 11
11:00 AM—12:00 PM (UTC)
Testimonials
Enjoyed watching all the in depth videos about Point Clouds as well as the theory that accompanied it! The bonus lecture with the command line interface was very cool as it can be abstracted into real-time processing with the software which I didn't think would be possible with just the GUI! If you would add anything new to the course, I would be fascinated to see that!
Saransh Chand
From Cloudcompare courseJ'ai particulièrement apprécié les cours théoriques avant la pratique
Kouamé Mondesir N'DRI
From QGIS courseI'm currently taking the "3D Point Cloud Masterclass | Lidar | CloudCompare" course on Udemy. The lessons are clear and practical, making complex concepts easy to understand. The hands-on exercises with CloudCompare are very helpful. Highly recommended for anyone interested in 3D point cloud processing!
Pathmila Jayasinghe
From Cloudcompare course