RECORD
Masked Autoencoders in 3D Point Cloud Representation Learning
Computer Vision and Pattern Recognition · Submitted 4 Jul 2022
abstract compiled · 2026-09-23 · no language-model review
The authors' main claim Comprehensive experiments demonstrate that the local features extracted by our MAE3D from point cloud patches are beneficial for downstream classification tasks, soundly outperforming state-of-the-art methods ($93.4\%$ and $86.2\%$ classification accuracy, respectively).
The compiled record
The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
Open the compiled record
The abstract, sentence by sentence
Transformer-based Self-supervised Representation Learning methods learn generic features from unlabeled datasets for providing useful network initialization parameters for downstream tasks. Recently, self-supervised learning based upon masking local surface patches for 3D point cloud data has been under-explored. In this paper, we propose masked Autoencoders in 3D point cloud representation learning (abbreviated as MAE3D), a novel autoencoding paradigm for self-supervised learning. We first split the input point cloud into patches and mask a portion of them, then use our Patch Embedding Module to extract the features of unmasked patches. Secondly, we employ patch-wise MAE3D Transformers to learn both local features of point cloud patches and high-level contextual relationships between patches and complete the latent representations of masked patches. We use our Point Cloud Reconstruction Module with multi-task loss to complete the incomplete point cloud as a result. We conduct self-supervised pre-training on ShapeNet55 with the point cloud completion pre-text task and fine-tune the pre-trained model on ModelNet40 and ScanObjectNN (PB\_T50\_RS, the hardest variant). Comprehensive experiments demonstrate that the local features extracted by our MAE3D from point cloud patches are beneficial for downstream classification tasks, soundly outperforming state-of-the-art methods ($93.4\%$ and $86.2\%$ classification accuracy, respectively).
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | Transformer-based Self-supervised Representation Learning methods learn generic features from unlabeled datasets for providing useful network initialization parameters for downstream tasks. | background 1.00 | 0.05 | 0.02 | asserted 1.00 |
| s1 | Recently, self-supervised learning based upon masking local surface patches for 3D point cloud data has been under-explored. | gap 0.99 | 0.07 | 0.02 | asserted 0.99 |
| s2 | In this paper, we propose masked Autoencoders in 3D point cloud representation learning (abbreviated as MAE3D), a novel autoencoding paradigm for self-supervised learning. | contribution 1.00 | 0.82 | 0.04 | asserted 0.98 |
| s3 | We first split the input point cloud into patches and mask a portion of them, then use our Patch Embedding Module to extract the features of unmasked patches. | method 1.00 | 0.91 | 0.02 | asserted 0.93 |
| s4 | Secondly, we employ patch-wise MAE3D Transformers to learn both local features of point cloud patches and high-level contextual relationships between patches and complete the latent representations of masked patches. | method 1.00 | 0.93 | 0.03 | asserted 0.97 |
| s5 | We use our Point Cloud Reconstruction Module with multi-task loss to complete the incomplete point cloud as a result. | method 1.00 | 0.94 | 0.04 | asserted 0.98 |
| s6 | We conduct self-supervised pre-training on ShapeNet55 with the point cloud completion pre-text task and fine-tune the pre-trained model on ModelNet40 and ScanObjectNN (PB\_T50\_RS, the hardest variant). | method 0.97 | 0.97 | 0.04 | asserted 0.90 |
| s7 | Comprehensive experiments demonstrate that the local features extracted by our MAE3D from point cloud patches are beneficial for downstream classification tasks, soundly outperforming state-of-the-art methods ($93.4\%$ and $86.2\%$ classification accuracy, respectively). | result 0.99 | 0.96 | 0.92 | asserted 1.00 |
What was examined, and what was not
- arXiv listing abstract v2
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
The paper
Original abstract
Transformer-based Self-supervised Representation Learning methods learn generic features from unlabeled datasets for providing useful network initialization parameters for downstream tasks. Recently, self-supervised learning based upon masking local surface patches for 3D point cloud data has been under-explored. In this paper, we propose masked Autoencoders in 3D point cloud representation learning (abbreviated as MAE3D), a novel autoencoding paradigm for self-supervised learning. We first split the input point cloud into patches and mask a portion of them, then use our Patch Embedding Module to extract the features of unmasked patches. Secondly, we employ patch-wise MAE3D Transformers to learn both local features of point cloud patches and high-level contextual relationships between patches and complete the latent representations of masked patches. We use our Point Cloud Reconstruction Module with multi-task loss to complete the incomplete point cloud as a result. We conduct self-supervised pre-training on ShapeNet55 with the point cloud completion pre-text task and fine-tune the pre-trained model on ModelNet40 and ScanObjectNN (PB\_T50\_RS, the hardest variant). Comprehensive experiments demonstrate that the local features extracted by our MAE3D from point cloud patches are beneficial for downstream classification tasks, soundly outperforming state-of-the-art methods ($93.4\%$ and $86.2\%$ classification accuracy, respectively).
Human review
References
How the literature uses it
6 arXiv papers cite this in 9 sentences
Papers citing it · 6 sentences without a confident label
Cites it as background 3 papers
-
More recent works adopt pre-trained CLIP [ref] for zero-shot 3D recognition [ref] , or introduce masked point modeling [ref] as strong 3D self-supervised learners.
Learning 3D Representations from 2D Pre-trained Models via Image-to-Point Masked Autoencoders · Related Work · 0.96
-
Recently, there has been a growing interest in exploring the shape description of the point cloud [ref] and different pre-training strategies [ref] .
ModelNet-O: A Large-Scale Synthetic Dataset for Occlusion-Aware Point Cloud Classification · Related Work · 0.96
-
Moreover, several works [ref] explore other point cloud architecture designs, including MLPs and transformers.
Point Cloud Mamba: Point Cloud Learning via State Space Model · Related Work · 1.00
Discussion
Know something about this paper? Leave a comment, or highlight any passage above to comment on it.