Points clouds are a common representation in applications, such as autonomous navigation, housekeeping robots and augmented/virtual reality. They provide reliable depth information that can be used to accurately localize objects and characterise their shapes. Unlike images, point clouds are sparse, have highly variable points density and require properly designed deep learning approaches.

Staggered plan

• What is point cloud?
• Why point clouds are challenging?
• Introduction to 3D Sensors
• Ways of representing point clouds for deep learning
• Overview of deep learning techniques
• Available datasets and benchmarks

Requirements for listeners:

• Understanding very basic concepts of math and deep learning.