Improved Point Cloud Representation via a Learnable Sort–Mix–Attend Mechanism
Open Access
Abstract
Recent years have seen remarkable progress in deep learning on 3D point clouds, with hierarchical architectures becoming standard. Most work has focused on developing increasingly complex operators, such as self-attention, while enhancing the representational capacity of efficient point-wise MLP-based backbones has received less attention. We address this issue by proposing a differentiable module that learns to impose a task-driven canonical structure on local point sets. Our proposed SMA (Sort–Mix–Attend) layer dynamically serializes a neighborhood by generating a geometric basis and using a differentiable sorting mechanism. This enables an efficient MLP-based network to model rich feature interactions, adaptively modulating features prior to the final symmetric aggregation function. We demonstrate that SMA effectively enhances standard backbones for 3D classification and segmentation. Specifically, integrating SMA into PointNeXt-S achieves an Overall Accuracy (OA) of 88.3% on the challenging ScanObjectNN dataset, an improvement of 0.6% over the baseline. Furthermore, it boosts the classic PointNet++ architecture by a significant 5.2% in OA. We also introduce a highly efficient SMA-Tiny variant that achieves 86.0% OA with only 0.3 M parameters, proving the structural superiority, computational cost-effectiveness, and practical significance of our method for real-world 3D perception tasks.Keywords
Funding Information
- Youth Program of the National Natural Science Foundation of China (62406208)
- International Cooperation Program of the Science and Technology Department of Sichuan Province (2025YFHZ0237)
This publication has 24 references indexed in Scilit:
- Advancements in point cloud data augmentation for deep learning: A surveyPattern Recognition, 2024
- Masked Autoencoders for Point Cloud Self-supervised Learning2022
- Geometric Transformer for Fast and Robust Point Cloud Registration2022
- Why Discard if You can Recycle?: A Recycling Max Pooling Module for 3D Point Cloud Analysis2022
- Stratified Transformer for 3D Point Cloud Segmentation2022
- PCT: Point cloud transformerComputational Visual Media, 2021
- RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Clouds2020
- 4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks2019
- 3D Semantic Parsing of Large-Scale Indoor Spaces2016
- Dritter Band: Analysis · Grundlagen der Mathematik · Physik Verschiedenes1935