Learn the fundamentals of Point Cloud Processing for 3D Object Detection, Segmentation and Classification. This online course is for individuals and companies who rapidly want to increase their 3D Perception skills without spending hours browsing and figuring out how to do it.
Course description
Discrete spatial datasets known as point clouds often lay the groundwork for decision-making applications. E.g., we can use such data as a reference for autonomous cars and robot’s navigation, as a layer for floor-plan’s creation and building’s construction, as a digital asset for environment modelling and incident prediction… Applications are numerous, and potentially increasing if we consider point clouds as digital reality assets. Yet, this expansion faces technical limitations mainly from the lack of experts in point cloud processing.
This course is a condensed hands-on lecture dedicated to providing you with focused content, immediately applied through an efficient fully-fledge open-source workflow. It treats point cloud data basics, engineering, semantization, structuration, analysis, visualisation and 3D modelling. The aim is to offer you the possibility to immediately integrate what you learnt in your personal or professional career.
After completing this course, you will be able to produce advanced automatic point cloud processing workflows of your own by using free and open-source software and efficient python code blocks. The provided principles or workflow can also be extended to other paid software (Trimble, Leica Cyclone, Faro Scene, Flyvast, …). You will be able to create advanced 3D automatic modelling workflows, master the full point cloud processing pipeline and produce stunning rendering to showcase your new expertise. All resources, articles, point clouds and 3D models used on this course are shared via a folder. Step-by-step tutorials are also provided as and videos commented as a “podcast” spirit.
This course is conceived to maximize the return on time invested and provide you with possibilities to upscale what you learnt and extend your expertise through additional focused formation programs. Its conception is based on years of R&D, and new knowledge-transmission techniques to access very high focus and limited time investment.
Key concepts covered include:
- Point Cloud Basics: How to start processing point cloud datasets from different sensors (LiDAR, Laser Scanner, Photogrammetry, ...)
- Point Cloud Engineering: Create advanced feature extraction and registration routines
- Point Cloud Semantization: Develop a pure semi-automatic segmentation procedure followed by classification using Machine Learning
- Point Cloud Analysis and Visualisation: Create robust qualitative and quantitative quality reports supported by unique 2D/3D visualisations
- Point Cloud Data Structure and Modelling: Apprehend 3D data structures (Octree, kd-tree) to accelerate processing and for 3D Modelling
- 3D Python automation: Put all 5 concepts together to create endless automatic procedures for advanced point cloud processing
Additionnal information
This course is designed for beginner to advanced user, to show the process of creation of highly targeted workflows. No knowledge is required. You only need a computer, with some free HDD space (100+ Go), some RAM (8 Go+) and preferably a decent dedicated GPU.
Also, future updates will include among others:
- Cutting-edge 3D Deep Learning (PointNet, PointNet++, RANDLANET, …)
- Apply Julia and Python, the best programming language for rapid results
- Develop a 3D segmentation workflow without user supervision
The additional materials are among the following:
- Articles (.pdf)
- Picture datasets (.jpeg)
- Mesh datasets (.obj, .mtl, .jpeg, .ply, .bin)
- Point Cloud datasets (.xyz, .las, .bin, .e57)
- Schematics (.jpeg, .pdfs)
- Formation Videos (online)
- Code and scripts (online, .py, .ipy)
These are optional but highly recommended to get full coverage of the learning potential. You will find downloadable content on each module presentation page. Videos are part of the modules themselves, and either constitute a topic or complement one.