PCT: Point cloud transformer

TL;DR

This paper introduces Point Cloud Transformer (PCT), a novel neural network framework leveraging Transformer architecture for effective point cloud learning, achieving state-of-the-art results in shape classification, segmentation, and normal estimation.

Contribution

The paper proposes PCT, a Transformer-based model tailored for point clouds, incorporating local context enhancement techniques like farthest point sampling and nearest neighbor search.

Findings

Achieves state-of-the-art performance on shape classification.

Excels in part segmentation tasks.

Provides accurate normal estimation results.

Abstract

The irregular domain and lack of ordering make it challenging to design deep neural networks for point cloud processing. This paper presents a novel framework named Point Cloud Transformer(PCT) for point cloud learning. PCT is based on Transformer, which achieves huge success in natural language processing and displays great potential in image processing. It is inherently permutation invariant for processing a sequence of points, making it well-suited for point cloud learning. To better capture local context within the point cloud, we enhance input embedding with the support of farthest point sampling and nearest neighbor search. Extensive experiments demonstrate that the PCT achieves the state-of-the-art performance on shape classification, part segmentation and normal estimation tasks.