This research paper presents a language-assisted approach to point cloud feature learning (LAST-PCL), aimed at improving 3D scene understanding. The approach incorporates knowledge from other modalities, like images and text, to assist in learning point cloud features. The method has been validated through extensive experiments, showing that it learns semantically meaningful point cloud features and achieves state-of-the-art performance in 3D semantic segmentation, 3D object detection, and 3D scene classification tasks.
Publication date: 19 Dec 2023
Project Page: https://github.com/yanmin-wu/LAST-PCL
Paper: https://arxiv.org/pdf/2312.11451
