Point Transformer for Shape Classification and Retrieval of 3D and ALS Roof PointClouds

TL;DR

This paper introduces Point Transformer, a fully attentional model for 3D point cloud classification and retrieval, demonstrating superior performance and robustness on large-scale datasets while being memory-efficient.

Contribution

It proposes a novel fully attentional Point Transformer model that replaces convolutions, achieving improved accuracy and robustness in 3D point cloud tasks.

Findings

Outperforms state-of-the-art on RoofN3D dataset

Achieves competitive results on ModelNet40

Shows high robustness to unseen point corruptions

Abstract

The success of deep learning methods led to significant breakthroughs in 3-D point cloud processing tasks with applications in remote sensing. Existing methods utilize convolutions that have some limitations, as they assume a uniform input distribution and cannot learn long-range dependencies. Recent works have shown that adding attention in conjunction with these methods improves performance. This raises a question: can attention layers completely replace convolutions? This paper proposes a fully attentional model - {\em Point Transformer}, for deriving a rich point cloud representation. The model's shape classification and retrieval performance are evaluated on a large-scale urban dataset - RoofN3D and a standard benchmark dataset ModelNet40. Extensive experiments are conducted to test the model's robustness to unseen point corruptions for analyzing its effectiveness on real datasets. The proposed method outperforms other state-of-the-art models in the RoofN3D dataset, gives competitive results in the ModelNet40 benchmark, and showcases high robustness to various unseen point corruptions. Furthermore, the model is highly memory and space efficient when compared to other methods.