OPTNet: Ordering Point Transformer Network for Post-disaster 3D Semantic Segmentation

TL;DR

OPTNet introduces a learnable point ordering mechanism for 3D semantic segmentation in post-disaster scenarios, enhancing efficiency and accuracy over fixed ordering methods.

Contribution

It proposes a novel self-supervised Point Sorter module that learns optimal point permutations to improve transformer-based segmentation.

Findings

Outperforms state-of-the-art baselines on 3DAeroRelief dataset.

Achieves higher accuracy and efficiency in post-disaster 3D semantic segmentation.

Demonstrates the effectiveness of learnable point ordering in complex scenes.

Abstract

Post-disaster damage assessment requires rapid and accurate semantic segmentation of 3D point clouds to identify critical infrastructure such as damaged buildings and roads. Early Point Transformers (e.g., PTv1, PTv2) relied on computationally expensive neighbor searching (k-NN) and Farthest Point Sampling (FPS). To improve efficiency, recent architectures like Point Transformer V3 (PTv3) adopted static serialization methods, such as Hilbert curves or Z-order, to organize unstructured points for window-based attention. However, these fixed orderings are not optimal for capturing the complex geometry of disaster scenes. In this paper, we propose OPTNet (Ordering Point Transformer Network), which introduces a learnable Point Sorter module. OPTNet utilizes a self-supervised ordering loss to dynamically predict an optimal permutation that maximizes the locality of the attention mechanism. We evaluate our method on the 3DAeroRelief dataset, significantly outperforming state-of-the-art baselines.