Point Cloud
A set of 3D points representing the surface or volume of an object or scene - alternative to meshes or voxels.
Think of it like
A cloud of dots in 3D space, each a measurement of a surface.
Example
LiDAR scan of a street: millions of points, each with (x, y, z) and optionally color or intensity. A object scan: 50K points on a statue.
How it actually works
Unstructured, no topology requirement (unlike meshes). Can be large (millions of points). Advantages: simple representation, no adjacency constraints, natural from sensors (LiDAR, depth cameras). Disadvantages: noisy, sparse, rendering/processing is non-trivial. ML on point clouds: PointNet (permutation invariant), PointNet++, Point Transformer. Applications: 3D object recognition, segmentation, tracking, autonomous driving.
For product teams
Autonomous vehicles, 3D imaging, scene understanding.
For engineers
Representation: (N, 3) array of x, y, z or (N, 6) with colors. Processing: PointNet-based architectures. Metric: Chamfer distance, Hausdorff distance.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome