Generative AI for 3-D Point Clouds

Abstract

Generative AI refers to algorithms and techniques designed to generate text, images, videos, or other data, typically in response to prompts. These algorithms leverage large generative models that learn the patterns and structures of the media data (text, images, or videos) provided during training and then generate new media data that have analogous characteristics. Much of the recent research has gone into applying generative AI for text and 2-D image data. However, generative AI for 3-D models, especially 3-D point cloud data (PCD), has compelling applications in virtual reality content generation, gaming, and product design and manufacturing, but it introduces a multitude of research challenges.