mlsanityData to 14 Oct 2026

Efficient Point Transformer with Dynamic Token Aggregating for LiDAR Point Cloud Processing

arXiv 2405.15827 ↗First posted 23 May 2024cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

Recently, LiDAR point cloud processing and analysis have made great progress due to the development of 3D Transformers. However, existing 3D Transformer methods usually are computationally expensive and inefficient due to their huge and redundant attention maps. They also tend to be slow due to requiring time-consuming point cloud sampling and grouping processes. To address these issues, we propose an efficient point TransFormer with Dynamic Token Aggregating (DTA-Former) for point cloud representation and processing. Firstly, we propose an efficient Learnable Token Sparsification (LTS) block, which considers both local and global semantic information for the adaptive selection of key tokens. Secondly, to achieve the feature aggregation for sparsified tokens, we present the first Dynamic Token Aggregating (DTA) block in the 3D Transformer paradigm, providing our model with strong aggregated features while preventing information loss. After that, a dual-attention Transformer-based Global Feature Enhancement (GFE) block is used to improve the representation capability of the model. Equipped with LTS, DTA, and GFE blocks, DTA-Former achieves excellent classification results via hierarchical feature learning. Lastly, a novel Iterative Token Reconstruction (ITR) block is introduced for dense prediction whereby the semantic features of tokens and their semantic relationships are gradually optimized during iterative reconstruction. Based on ITR, we propose a new W-net architecture, which is more suitable for Transformer-based feature learning than the common U-net design.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 24 of its 45 references that are in the map. First posted on arXiv 23 May 2024.

Builds on

24 of 45 references.

References · 24

  1. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,139
  2. 2016PointNet: Deep Learning on Point Sets for 3D Classification and Segmentationcited by 5,248
  3. 2017PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Spacecited by 4,401
  4. 2016Categorical Reparameterization with Gumbel-Softmaxcited by 2,739
  5. 2018Dynamic Graph CNN for Learning on Point Cloudscited by 2,223
  6. 2019Transformer-XL: Attentive Language Models Beyond a Fixed-Length Contextcited by 1,850
  7. 2020Point Transformercited by 1,470
  8. 2018PointCNN: Convolution On X-Transformed Pointscited by 764
  9. 2020PCT: Point cloud transformercited by 732
  10. 2019RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Cloudscited by 585
  11. 2019DeepGCNs: Can GCNs Go as Deep as CNNs?cited by 527
  12. 2017Large-scale Point Cloud Semantic Segmentation with Superpoint Graphscited by 385
12 more
  1. 2019Relation-Shape Convolutional Neural Network for Point Cloud Analysiscited by 346
  2. 2020PointASNL: Robust Point Clouds Processing using Nonlocal Neural Networks with Adaptive Samplingcited by 239
  3. 2023DSVT: Dynamic Sparse Voxel Transformer with Rotated Setscited by 122
  4. 2022Superpoint Transformer for 3D Scene Instance Segmentationcited by 116
  5. 2022Efficient Transformers with Dynamic Token Poolingcited by 76
  6. 2023FlatFormer: Flattened Window Attention for Efficient Point Cloud Transformercited by 73
  7. 2019DeepGCNs: Making GCNs Go as Deep as CNNscited by 72
  8. 2021Adaptive Graph Convolution for Point Cloud Analysiscited by 60
  9. 2023Efficient 3D Semantic Segmentation with Superpoint Transformercited by 57
  10. 2021PatchFormer: An Efficient Point Transformer with Patch Attentioncited by 43
  11. 2023Attention-based Point Cloud Edge Samplingcited by 37
  12. 2023Point Cloud Classification Using Content-based Transformer via Clustering in Feature Spacecited by 19

Built on by

Not cited yet.