Part 1

3D Data Series: Intro to Point Cloud Data

In Part 1, we will explore:

• What is point cloud data
• Obtaining point cloud data
• Labelling point cloud data
• Applications to machine learning

In Part 2, we will explore:

• Preprocessing point cloud data
• Visualizing point cloud data

Image from pinterest

What is Point Cloud Data

I first heard the term “point cloud” while reading an Uber Engineering article on Aerial ridesharing. The article introduces 3D tiles, a structured data format for rendering complex 3D models — including point cloud data to seamlessly visualize high-resolution maps and dynamic urban landscapes that support precise flight path planning, obstacle avoidance, and real-time situational awareness, all crucial for aerial ridesharing operations.

Delving deeper into these 3d models, point clouds are volumetric data that contains information such as 3d coordinates of the object’s features (either in global or reference frame), Intensity (strength of the return signal), GPS timestamps and textures depending on the type of scanning technology used to capture the details. So, if you are scanning a building, each point would represent a real point on the wall…