# LighTN: Light-weight Transformer Network for Performance-overhead Tradeoff in Point Cloud Downsampling

arXiv: https://arxiv.org/abs/2202.06263 · HTML: https://mlsanity.com/paper/2202.06263/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- Builds on 33 of its 56 references that are in the map.
- First posted on arXiv 13 Feb 2022.

## Builds on (most cited first, 30 of 33)

- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation](https://mlsanity.com/paper/1612.00593/) (2016)
- [PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space](https://mlsanity.com/paper/1706.02413/) (2017)
- [Dynamic Graph CNN for Learning on Point Clouds](https://mlsanity.com/paper/1801.07829/) (2018)
- [Point Transformer](https://mlsanity.com/paper/2011.00931/) (2020)
- [VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection](https://mlsanity.com/paper/1711.06396/) (2017)
- [PointCNN: Convolution On X-Transformed Points](https://mlsanity.com/paper/1801.07791/) (2018)
- [PCT: Point cloud transformer](https://mlsanity.com/paper/2012.09688/) (2020)
- [Transformer in Transformer](https://mlsanity.com/paper/2103.00112/) (2021)
- [RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Clouds](https://mlsanity.com/paper/1911.11236/) (2019)
- [Relation-Shape Convolutional Neural Network for Point Cloud Analysis](https://mlsanity.com/paper/1904.07601/) (2019)
- [PointASNL: Robust Point Clouds Processing using Nonlocal Neural Networks with Adaptive Sampling](https://mlsanity.com/paper/2003.00492/) (2020)
- [Mobile-Former: Bridging MobileNet and Transformer](https://mlsanity.com/paper/2108.05895/) (2021)
- [End-to-End Multi-View Fusion for 3D Object Detection in LiDAR Point Clouds](https://mlsanity.com/paper/1910.06528/) (2019)
- [The Perfect Match: 3D Point Cloud Matching with Smoothed Densities](https://mlsanity.com/paper/1811.06879/) (2018)
- [MeteorNet: Deep Learning on Dynamic 3D Point Cloud Sequences](https://mlsanity.com/paper/1910.09165/) (2019)
- [Local Spectral Graph Convolution for Point Set Feature Learning](https://mlsanity.com/paper/1803.05827/) (2018)
- [EC-Net: an Edge-aware Point set Consolidation Network](https://mlsanity.com/paper/1807.06010/) (2018)
- [P2B: Point-to-Box Network for 3D Object Tracking in Point Clouds](https://mlsanity.com/paper/2005.13888/) (2020)
- [Recurrent Slice Networks for 3D Segmentation of Point Clouds](https://mlsanity.com/paper/1802.04402/) (2018)
- [SampleNet: Differentiable Point Cloud Sampling](https://mlsanity.com/paper/1912.03663/) (2019)
- [PCRNet: Point Cloud Registration Network using PointNet Encoding](https://mlsanity.com/paper/1908.07906/) (2019)
- [A Rotation-Invariant Framework for Deep Point Cloud Analysis](https://mlsanity.com/paper/2003.07238/) (2020)
- [Pointfilter: Point Cloud Filtering via Encoder-Decoder Modeling](https://mlsanity.com/paper/2002.05968/) (2020)
- [VPFNet: Improving 3D Object Detection with Virtual Point based LiDAR and Stereo Data Fusion](https://mlsanity.com/paper/2111.14382/) (2021)
- [End-to-End Learning Local Multi-view Descriptors for 3D Point Clouds](https://mlsanity.com/paper/2003.05855/) (2020)
- [Adaptive Hierarchical Down-Sampling for Point Cloud Classification](https://mlsanity.com/paper/1904.08506/) (2019)
- [A Battle of Network Structures: An Empirical Study of CNN, Transformer, and MLP](https://mlsanity.com/paper/2108.13002/) (2021)
- [MOPS-Net: A Matrix Optimization-driven Network forTask-Oriented 3D Point Cloud Downsampling](https://mlsanity.com/paper/2005.00383/) (2020)

## Built on by (most cited first, 8 of 8)

- [Attention-based Point Cloud Edge Sampling](https://mlsanity.com/paper/2302.14673/) (2023)
- [Transformers in 3D Point Clouds: A Survey](https://mlsanity.com/paper/2205.07417/) (2022)
- [Self-Attentive Pooling for Efficient Deep Learning](https://mlsanity.com/paper/2209.07659/) (2022)
- [A Correct-and-Certify Approach to Self-Supervise Object Pose Estimators via Ensemble Self-Training](https://mlsanity.com/paper/2302.06019/) (2023)
- [Edge Aware Learning for 3D Point Cloud](https://mlsanity.com/paper/2309.13472/) (2023)
- [Precise Workcell Sketching from Point Clouds Using an AR Toolbox](https://mlsanity.com/paper/2410.00479/) (2024)
- [SAMBLE: Shape-Specific Point Cloud Sampling for an Optimal Trade-Off Between Local Detail and Global Uniformity](https://mlsanity.com/paper/2504.19581/) (2025)
- [REPS: Reconstruction-based Point Cloud Sampling](https://mlsanity.com/paper/2403.05047/) (2024)

## Abstract (arXiv)

Compared with traditional task-irrelevant downsampling methods, task-oriented neural networks have shown improved performance in point cloud downsampling range. Recently, Transformer family of networks has shown a more powerful learning capacity in visual tasks. However, Transformer-based architectures potentially consume too many resources which are usually worthless for low overhead task networks in downsampling range. This paper proposes a novel light-weight Transformer network (LighTN) for task-oriented point cloud downsampling, as an end-to-end and plug-and-play solution. In LighTN, a single-head self-correlation module is presented to extract refined global contextual features, where three projection matrices are simultaneously eliminated to save resource overhead, and the output of symmetric matrix satisfies the permutation invariant. Then, we design a novel downsampling loss function to guide LighTN focuses on critical point cloud regions with more uniform distribution and prominent points coverage. Furthermore, We introduce a feed-forward network scaling mechanism to enhance the learnable capacity of LighTN according to the expand-reduce strategy. The result of extensive experiments on classification and registration tasks demonstrates LighTN can achieve state-of-the-art performance with limited resource overhead.