PU-T2M: Heterogeneous Knowledge Distilling Network From Transformer to Mamba for Lightweight Point Cloud Upsampling
- 1 November 2026
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 72 (4) , 8070-8087
Abstract
Three-dimensional point clouds have become a core sensing modality in modern consumer electronics systems, particularly in intelligent vehicles equipped with low-power LiDAR and depth cameras. However, the point clouds captured by these consumer-grade sensors are often sparse and noisy due to complex environment and sensor constraints. Such degradation adversely affects downstream perception tasks, including real-time scene reconstruction, object understanding, and navigation stability. Point Cloud Upsampling (PCU) is therefore essential for restoring structural fidelity and enabling reliable three dimensions (3D) environment modeling in resource-constrained edge-to-cloud automotive systems. To overcome these issues, we propose a novel lightweight PCU architecture based on Heterogeneous Knowledge Distilling Network from Transformer to Mamba, termed PU-T2M. This workflow achieves state-of-the-art (SOTA) performance while being computationally efficient, making it suitable for edge devices with limited resources. Specifically, our network distills dark knowledge from Transformer to Mamba, enabling the model to learn long-range global context while preserving local geometric structure. We incorporate Mamba and knowledge distillation into the PCU task to achieve a more efficient and effective modeling framework. Specifically, we design a dual-alignment distillation loss that facilitates knowledge transfer while enforcing both explicit alignment in point distributions and implicit alignment in feature representations. In addition, we develop a Global Context Preserver module and a Local Geometry Preserver module to strengthen the Transformer’s global contextual modeling capability and enhance Mamba’s capacity for fine-grained local geometric learning. Extensive experiments conducted on four benchmark datasets demonstrate that PU-T2M attains state-of-the-art performance while maintaining a compact parameter scale and strong robustness. The proposed approach enables accurate and computationally efficient point cloud reconstruction, making it particularly suitable for industrial metrology and high-precision instrument-level quality inspection scenarios.Keywords
Funding Information
- Natural Science Foundation of China (62471263)
- “Science and Technology Innovation Yongjiang 2035” Key R&D Plan Project in Ningbo (2025Z043)
- Ningbo Natural Science Foundation of China (2025J113)
This publication has 60 references indexed in Scilit:
- PU-MG: Mutual guidance framework for Point Cloud UpsamplingDisplays, 2025
- LiGAPU: A LiDAR point cloud upsampling network for multiple complex scenesPattern Recognition, 2025
- QCTKD-PU: Quantum Convolutional Transformer with Knowledge Distillation for Efficient and Robust Point Cloud Upsampling2025
- A Noising-Denoising Framework for Point Cloud Upsampling via Normalizing FlowsPattern Recognition, 2023
- Point Cloud Upsampling via Cascaded Refinement Network2023
- CAVE-VR and Unity Game Engine for Visualizing City Scale 3D Meshes2022
- PCT: Point cloud transformerComputational Visual Media, 2021
- PUGeo-Net: A Geometry-Centric Network for 3D Point Cloud Upsampling2020
- Commodity 3D display based on point cloud reconstruction2018
- The ball-pivoting algorithm for surface reconstructionIEEE Transactions on Visualization and Computer Graphics, 1999