Pith. sign in

REVIEW 3 major objections 6 minor 43 references

HFBRI-MAE: Handcrafted Feature Based Rotation-Invariant Masked Autoencoder for 3D Point Cloud Analysis

T0 review · 3 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Rotation-invariant point-cloud MAE holds above 91.3% accuracy.

desk verdict Useful MAE+handcrafted-feature combination, but the formal rotation-invariance proof has an unaddressed LRA sign ambiguity. read the letter →

arxiv 2504.14132 v1 pith:OK6MTWY5 submitted 2025-04-19 cs.CV

Xuanhua Yin , Dingxin Zhang , Jianhui Yu , Weidong Cai This is my paper · ORCID
classification cs.CV
keywords 3Dpointcloudrotationinvariancemaskedautoencoderself-supervisedlearninghandcraftedfeatureslocalreferenceaxisclassificationfew-shot
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Self-supervised masked autoencoders (MAEs) for 3D point clouds lose accuracy when objects appear in arbitrary orientations, because both token and position embeddings encode absolute coordinates. HFBRI-MAE claims to fix this by feeding the network only handcrafted rotation-invariant features, using local angle and distance features for tokens and ball-based global features for positions, and by asking the decoder to reconstruct a canonically aligned version of the input rather than the original rotated shape. The paper reports that this design keeps classification accuracy above 91.3% across all tested rotation settings on ModelNet40 and beats prior rotation-invariant and SSL methods on classification, part segmentation, and few-shot learning. If correct, it shows that rotation invariance for point-cloud SSL does not require learned equivariant layers or local reference frames; handcrafted geometric features and an aligned reconstruction target are enough.

What carries the argument

The load-bearing object is the Rotation-Invariant Handcrafted Feature set (RIHF), made of two pieces. RILF captures fine-grained local geometry in each patch: the distance from reference point to neighbor, three reference-point angles, and four inter-neighbor angles, all measured relative to a Local Reference Axis (LRA) defined as the smallest eigenvector of the local covariance; points are projected onto the tangent plane and ordered clockwise from the farthest point to make the angles well-defined. RIGF encodes global layout with a neighborhood ball centered on each patch reference point, giving three distances (origin-to-p, p-to-centroid, centroid-to-boundary) and two angles. These features replace both token and position embeddings, while the reconstruction head predicts aligned coordinates P_align under a Chamfer-distance loss; together this machinery is what the paper claims eliminates rotational dependencies while keeping MAE reconstruction well-posed.

What would settle it

Take a fixed local patch, compute the smallest eigenvector of its covariance, rotate the patch so that eigenvector maps to its negative, and compare the eight RILF values; any difference between the original and rotated descriptors would falsify the claimed rotation invariance in Eq. (2).

Watch

Extended reading notes

Core claim

On its own terms, the paper establishes that a masked autoencoder can be made rotation-invariant in practice by replacing coordinate-based inputs with rotation-invariant handcrafted features and by changing the reconstruction target. Formally, it requires RIHF(P) = RIHF(RP) for every R in SO(3) and redefines the decoder objective as g(RIHF(RP)) leads to P_align, where P_align is the input after canonical coordinate normalization; this removes the ambiguity a rotation-invariant encoder creates for reconstruction. The handcrafted representation combines RILF, an eight-dimensional local descriptor of point distances and angles built around a Local Reference Axis, with RIGF, a five-dimensional global descriptor of ball-based distances and angles used for position embedding. Empirically the paper claims classification accuracies of 91.3 to 91.7 percent on ModelNet40 across the A/A, A/R, Z/Z, Z/R, and R/R settings, a C-mIoU of 79.5 percent on ShapeNetPart under Z/R, and consistent wins on 5-way and 10-way few-shot classification, concluding that handcrafted features plus aligned reconstruction targets are sufficient for practical rotation invariance in MAE-based point cloud analysis.

Load-bearing premise

The construction assumes the local reference direction, the smallest eigenvector of a patch's covariance matrix, keeps a consistent orientation under rotation, but an eigenvector can flip sign, which would reverse the clockwise ordering and change the angle features.

Editorial extensions

If this is right

  • A masked autoencoder can be made rotation-invariant without equivariant layers or learned pose alignment, so pretraining on unaligned real-world scans becomes feasible.
  • On ModelNet40, a single finetuned model maintains 91.3 to 91.7 percent accuracy across aligned, z-rotated, and fully random rotations, so deployment no longer requires canonicalizing inputs.
  • On ScanObjectNN-BG, accuracy holds at 89.4 to 90.5 percent under random rotations, indicating the invariance transfers to noisy, occluded real-world data.
  • The aligned-reconstruction objective gives a well-posed pretraining signal: the decoder learns to rebuild normalized geometry even though its input features carry no orientation information.
  • Few-shot 5-way and 10-way classification on ModelNet40 improves over prior rotation-invariant SSL methods in every rotation setting, suggesting the learned features transfer better to limited-label regimes.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper criticizes PCA sign ambiguity but never resolves its own Local Reference Axis sign ambiguity: if the smallest eigenvector flips sign, the clockwise ordering reverses and the angle features change, so strict invariance in Eq. (2) is not guaranteed; a sign-canonicalization heuristic is a direct, testable fix.
  • Because RIGF's first distance is measured from the origin, the method is rotation-invariant but not translation-invariant; applying it to arbitrary scenes would likely need a global reference shift or a translation-invariant global descriptor.
  • The A/A gain over prior MAEs may partly reflect the richer handcrafted local geometry rather than rotation handling; ablating HFBRI-MAE against a non-invariant MAE using the same RILF would quantify how much of the gain comes from invariance.
  • The paper mentions adaptive masking only as future work, but it is immediately testable: mask patches by local geometric complexity and see whether the 89.6 percent SVM accuracy on ModelNet40 improves.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper proposes HFBRI-MAE, a masked autoencoder for 3D point cloud analysis that replaces raw coordinates with handcrafted rotation-invariant local features (RILF) and global features (RIGF) for token and position embeddings, and reconstructs a canonically aligned point cloud to resolve orientation ambiguity. The authors report experiments on ModelNet40, ScanObjectNN, ShapeNetPart, ModelNetFewShot, and (in the appendix) OmniObject3D, claiming state-of-the-art accuracy under rotated settings, with finetuned classification accuracies above 91.3% on ModelNet40 across all tested rotation configurations.

Significance. If the rotation-invariance claim is correct, the paper would provide a practical demonstration that handcrafted geometry features plus an aligned reconstruction target are sufficient to make masked autoencoders robust to arbitrary 3D rotations, a problem of clear practical relevance. The paper's strengths include the breadth of experiments (three tasks, multiple benchmarks, and nine rotation settings in the appendix), the sensible design of an aligned reconstruction target to avoid orientation ambiguity, and ablations of mask ratio and feature components. The empirical R/R accuracy of about 91.5% on ModelNet40 is a concrete, falsifiable result. However, the theoretical foundation of the invariance claim is not established, and the headline performance claim is overstated relative to the reported numbers.

major comments (3)
  1. [Section III-C and Appendix A, Eq. (2)] The central invariance claim, Eq. (2), is not proven because the Local Reference Axis (LRA) is defined as the smallest eigenvector of the local covariance matrix and is therefore determined only up to sign. The signed angle features alpha2 and beta2, as well as the clockwise ordering of projected points, depend on the orientation of this axis; a sign flip reverses the ordering and flips the signs of these angles, changing RILF. The paper criticizes PCA sign ambiguity in Section II-B but does not fix the sign of its own LRA, and for a single eigenvector no continuous rotation-invariant sign convention exists. Consequently there are rotations R for which RIHF(P) != RIHF(RP), directly contradicting Eq. (2). Since Eq. (3) and the entire reconstruction objective rely on Eq. (2), this is a load-bearing issue rather than a cosmetic one.
  2. [Abstract and Tables I, II, IV, V] The abstract and conclusion state that HFBRI-MAE "consistently outperforms existing methods," but the reported tables contradict this for aligned (A/A) settings. In Table I, PointM2AE, PointGPT, and MaskSurf achieve 92.6%, 92.6%, and 92.4% on A/A, respectively, while HFBRI-MAE (Finetune) achieves 91.3%. In Table II, PointGPT reaches 91.6% on A/A versus HFBRI-MAE's 90.1%. In the few-shot tables, PointMAE and PointM2AE exceed HFBRI-MAE on A/A. The claim should be restricted to rotated settings (A/R, Z/R, R/R) and to comparisons against rotation-invariant methods, where the reported results are indeed strong.
  3. [Section IV-A and Section III-E] The training protocol for rotated settings is underspecified. Section IV-A defines X/Y with X denoting rotation applied during training, "including both pretraining and subsequent finetuning or SVM training," but Section III-E states that the pretraining dataset consists of pre-aligned point clouds used directly as Palign. No description is given of how Align(RP) is computed to serve as the reconstruction target if pretraining is performed on rotated data. This makes it impossible to reproduce the R/R and Z/R pretraining experiments and leaves the relationship between the aligned reconstruction target and rotated pretraining inputs ambiguous.
minor comments (6)
  1. [Section IV-C and Appendix D] There are numerical and labeling conflicts: the text in Section IV-C says a mask ratio of 0.6 gives 89.6%, but Table VI shows 89.7%; the text in Appendix D says Table X shows 89.6% for the proposed global features, but the table shows "Ours" at 91.4%; and the heading "Completed Classfication Results" refers to Tables XI-XIII while the actual tables are numbered VIII-X.
  2. [Section III-C and Appendix A] The main text defines the point m as the centroid of the neighborhood ball, while Appendix A calls m the geometric median. This distinction matters for RIGF's dpm and dsm features; please clarify which one is used and provide the exact computation.
  3. [Section IV-C and Table VII] There is a typo in Table VII's header ("angels" for "angles"), and the phrase "thes-to-m" in Section III-C appears to be a typo for "s-to-m."
  4. [Eq. (3)] Equation (3) uses an arrow in a nonstandard way; it should be made explicit that the equation states g(RIHF(P)) = g(RIHF(RP)) reconstructs Palign, i.e., the decoder maps invariant features to the aligned coordinate frame.
  5. [All experiment tables] No error bars, standard deviations, or multiple-run averages are reported. Given that several reported advantages over PaRot and RIConv++ are 0.2–0.5 percentage points, the authors should state whether these differences are significant or provide repeated-run statistics.
  6. [Appendix D] The appendix introduces OmniObject3D without describing its preparation, number of classes, or evaluation protocol; a brief description should be added for completeness.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: HFBRI-MAE's rotation-invariant handcrafted features and aligned reconstruction target are independently constructed; self-citations to PaRot are baselines, not load-bearing premises.

full rationale

HFBRI-MAE is an empirical architecture paper whose rotation-invariance claim rests on the explicit construction of handcrafted features (Eqs. 4 and 5) rather than on a fitted parameter or a self-citation chain. The reconstruction target is defined as the aligned input point cloud (Eq. 3, Eq. 9), which is a standard self-supervised objective: no quantity is fitted to a subset of data and then reported as a prediction of that same subset. The paper does compare against PaRot [24], a baseline from the same research group, and cites other works by the same authors for related context, but these citations are not load-bearing for the central claim, which is evaluated against external benchmarks (ModelNet40, ScanObjectNN, ShapeNetPart) and against many independent methods. The LRA sign ambiguity identified by a skeptical reader is a correctness and stability concern about Eq. (2), not a circularity: the invariance claim is asserted by construction, and if the smallest-eigenvector sign is not stabilized, the feature would fail to be invariant under certain rotations. That would be a mathematical flaw, not a reduction of the paper's output to its input. No step in the paper reduces, by its own equations or by self-citation, to its own inputs, so no circular step is exhibited.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

No new physical entities or fitted scientific constants appear. The handcrafted features (RILF, RIGF) are formulas over the input, and the listed hyperparameters were tuned on validation splits via ablations in the paper.

free parameters (4)
  • Mask ratio = 0.6
    Selected via ablation in Table VI; directly affects pretraining difficulty and downstream accuracy.
  • Number of patches = 256
    Selected via ablation in Table VIII; balances local coverage and redundancy.
  • Points per patch = 64
    Selected via ablation in Table VIII; controls local geometric detail.
  • Token and position embedding dimension = 384
    Architecture hyperparameter for the transformer, not a fitted scientific constant.
assumptions (4)
  • domain assumption The smallest eigenvector of the local covariance matrix provides a stable local reference axis across rotations.
    Used in RILF (Section III-C) to define LRA and angle features; valid if neighborhoods are non-degenerate, but eigenvector sign is not unique.
  • domain assumption The clockwise ordering of projected points around the reference point is consistent under arbitrary rotations.
    Invoked in Section III-C for inter-neighbor angle features; depends on a fixed orientation of the tangent plane and the LRA sign.
  • domain assumption Input point clouds are centered at the origin and not translated, so RIGF's origin-based distances are meaningful.
    RIGF (Section III-C and Appendix A) uses the coordinate origin; Eq. (1) only covers rotation, not translation.
  • domain assumption Pretraining data are already aligned, so the aligned input itself can serve as the canonical reconstruction target.
    Section III-E states the pretraining dataset consists of pre-aligned point clouds used directly as Palign; this avoids external canonicalization but limits the method to centered, orientation-consistent data at pretraining time.

how reviews work

0 comments
Cite this review

Pith. "Pith review of HFBRI-MAE: Handcrafted Feature Based Rotation-Invariant Masked Autoencoder for 3D Point Cloud Analysis." pith.science (2026). https://pith.science/paper/OK6MTWY5

@misc{pith2026250414132,
  author       = {Pith},
  title        = {Pith review of: HFBRI-MAE: Handcrafted Feature Based Rotation-Invariant Masked Autoencoder for 3D Point Cloud Analysis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OK6MTWY5}},
  note         = {Machine review of arXiv:2504.14132}
}
read the original abstract

Self-supervised learning (SSL) has demonstrated remarkable success in 3D point cloud analysis, particularly through masked autoencoders (MAEs). However, existing MAE-based methods lack rotation invariance, leading to significant performance degradation when processing arbitrarily rotated point clouds in real-world scenarios. To address this limitation, we introduce Handcrafted Feature-Based Rotation-Invariant Masked Autoencoder (HFBRI-MAE), a novel framework that refines the MAE design with rotation-invariant handcrafted features to ensure stable feature learning across different orientations. By leveraging both rotation-invariant local and global features for token embedding and position embedding, HFBRI-MAE effectively eliminates rotational dependencies while preserving rich geometric structures. Additionally, we redefine the reconstruction target to a canonically aligned version of the input, mitigating rotational ambiguities. Extensive experiments on ModelNet40, ScanObjectNN, and ShapeNetPart demonstrate that HFBRI-MAE consistently outperforms existing methods in object classification, segmentation, and few-shot learning, highlighting its robustness and strong generalization ability in real-world 3D applications.

Figures

Figures reproduced from arXiv: 2504.14132 by the authors.

Figure 1
Figure 1. Comparison between standard MAE and HFBRI-MAE [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Architecture of the proposed HFBRI-MAE framework. The input point cloud is divided into patches using FPS and [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Visualization of distance features (dpxi) and angle relationships with reference point (α0, α1, α2) in RILF. embedding tokens. The visible patches are first converted to RIHF, then converted to embeddings through designed embedding networks. Although adopting a fixed masking strategy facilitates training stability and implementation sim￾plicity, it may limit the model’s adaptability to varying input structures. In c… view at source ↗
Figures from the paper (4 more)
Figure 5
Figure 5. Figure 5: Visualisation of the RIGF (dp, dpm, dsm, α, β) con￾struction process using the neighborhood ball, which is cen￾tered at the reference point p with a radius r defined by the distance to the farthest neighboring point. Embedding. The embedding process transforms RIHF int…
Figure 6
Figure 6. Figure 6: Visualization of part segmentation results on the [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Embedding architecture overview. The left branch [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 9
Figure 9. Figure 9: Architecture of the segmentation head. The diagram [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 37 canonical work pages

  1. [1]

    Deep learning for 3d point clouds: A survey,

    Y . Guo, H. Wang, Q. Hu, H. Liu, L. Liu, and M. Bennamoun, “Deep learning for 3d point clouds: A survey,” in IEEE Transactions on Pattern Analysis and Machine Intelligence , 2020, 43(12), pp. 4338-4364

    work page 2020

  2. [2]

    Deep learning for image and point cloud fusion in autonomous driving: A review,

    Y . Cui et al., “Deep learning for image and point cloud fusion in autonomous driving: A review,” in IEEE Transactions on Intelligent Transportation Systems, 2021, 23(2), pp. 722–739

    work page 2021

  3. [3]

    A morphing-based 3D point cloud reconstruction framework for medical image processing,

    Q. Cheng, P. Sun, C. Yang, Y . Yang, and P. X. Liu, “A morphing-based 3D point cloud reconstruction framework for medical image processing,” in Computer Methods and Programs in Biomedicine , 2020, 193, p. 105495

    work page 2020

  4. [4]

    3d medical point transformer: Introducing convolution to attention networks for medical point cloud analysis,

    J, Yu et al. “3d medical point transformer: Introducing convolution to attention networks for medical point cloud analysis, ” arXiv preprint arXiv:2112.04863 (2021)

    arXiv 2021

  5. [5]

    Robotics dexterous grasping: The methods based on point cloud and deep learn- ing,

    H. Duan, P. Wang, Y . Huang, G. Xu, W. Wei, and X. Shen, “Robotics dexterous grasping: The methods based on point cloud and deep learn- ing,” in Frontiers in Neurorobotics, 2021, 15, p. 658280

    work page 2021

  6. [6]

    Point- Contrast: Unsupervised pre-training for 3D point cloud understanding,

    S. Xie, G. Jiatao, D. Guo, C. R. Qi, L. Guibas, and O. Litany, “Point- Contrast: Unsupervised pre-training for 3D point cloud understanding,” in ECCV, 2020, pp. 574–591

    work page 2020

  7. [7]

    CrossPoint: Self-supervised cross-modal contrastive learning for 3D point cloud understanding,

    M. Afham, I. Dissanayake, D. Dissanayake, A. Dharmasiri, K. Thi- lakarathna, and R. Rodrigo, “CrossPoint: Self-supervised cross-modal contrastive learning for 3D point cloud understanding,” in CVPR, 2022, pp. 9902–9912

    work page 2022

  8. [8]

    Exploring geometry-aware contrast and clustering harmonization for self-supervised 3D object detection,

    H. Liang et al., “Exploring geometry-aware contrast and clustering harmonization for self-supervised 3D object detection,” in ICCV, 2021, pp. 3293–3302

    work page 2021

Show all 43 references
  1. [9]

    FoldingNet: Point cloud auto- encoder via deep grid deformation,

    Y . Yang, C. Feng, Y . Shen, and D. Tian, “FoldingNet: Point cloud auto- encoder via deep grid deformation,” in CVPR, 2018, pp. 206–215

    2018

  2. [10]

    Progressive seed generation auto-encoder for unsupervised point cloud learning,

    J. Yang, P. Ahn, D. Kim, H. Lee, and J. Kim, “Progressive seed generation auto-encoder for unsupervised point cloud learning,” inICCV, 2021, pp. 6413–6422

    2021

  3. [11]

    Masked autoencoders for point cloud self-supervised learning,

    Y . Pang, W. Wang, F. E. Tay, W. Liu, Y . Tian, and L. Yuan, “Masked autoencoders for point cloud self-supervised learning,” in ECCV, 2022, pp. 604–621

    2022

  4. [12]

    Point-M2AE: Multi-scale masked autoencoders for hierarchical point cloud pre-training,

    R. Zhang et al., “Point-M2AE: Multi-scale masked autoencoders for hierarchical point cloud pre-training,” in NeurIPS, 2022, 35, pp. 27061– 27074

    2022

  5. [13]

    MaskLRF: Self-supervised Pretraining via Masked Autoen- coding of Local Reference Frames for Rotation-invariant 3D Point Set Analysis,

    T. Furuya, “MaskLRF: Self-supervised Pretraining via Masked Autoen- coding of Local Reference Frames for Rotation-invariant 3D Point Set Analysis,” in IEEE Access , 2024

    2024

  6. [14]

    RI-MAE: Rotation-Invariant Masked AutoEncoders for Self-Supervised Point Cloud Representation Learning,

    K. Su et al., “RI-MAE: Rotation-Invariant Masked AutoEncoders for Self-Supervised Point Cloud Representation Learning,” in arXiv preprint, 2024, arXiv:2409.00353

    2024 arXiv

  7. [15]

    Masked surfel prediction for self-supervised point cloud learning,

    Y . Zhang, J. Lin, C. He, Y . Chen, K. Jia, and L. Zhang, “Masked surfel prediction for self-supervised point cloud learning,” in arXiv preprint , 2022, arXiv:2207.03111

    2022 arXiv

  8. [16]

    Point- GPT: Auto-regressively generative pre-training from point clouds,

    G. Chen, M. Wang, Y . Yang, K. Yu, L. Yuan, and Y . Yue, “Point- GPT: Auto-regressively generative pre-training from point clouds,” in NeurIPS, 2024, vol. 36

    2024

  9. [17]

    Enhancing Robustness to Noise Corruption for Point Cloud Recognition via Spatial Sorting and Set-Mixing Aggregation Module,

    D. Zhang, J. Yu, T. Xue, C. Zhang, D. Liu, and W. Cai, “Enhancing Robustness to Noise Corruption for Point Cloud Recognition via Spatial Sorting and Set-Mixing Aggregation Module,” in ACCV, 2024, pp. 4205- 4222

    2024

  10. [18]

    Rotation invariant convolutions for 3D point clouds deep learning,

    Z. Zhang, B. S. Hua, W. David, Rosen, and S. K. Yeung, “Rotation invariant convolutions for 3D point clouds deep learning,” in 3DV, 2019, pp. 204–213

    2019

  11. [19]

    RIConv++: Effective rotation in- variant convolutions for 3D point clouds deep learning,

    Z. Zhang, B. S. Hua, and S. K. Yeung, “RIConv++: Effective rotation in- variant convolutions for 3D point clouds deep learning,” in International Journal of Computer Vision , 2022, 130(5), pp. 1228–1243

    2022

  12. [20]

    Global context aware convolutions for 3D point cloud understanding,

    Z. Zhang, B. S. Hua, W. Chen, Y . Tian, and S. K. Yeung, “Global context aware convolutions for 3D point cloud understanding,” in 3DV, 2020, pp. 210–219

    2020

  13. [21]

    Rotation invariant point cloud analysis: Where local geometry meets global topology,

    C. Zhao, J. Yang, X. Xiong, A. Zhu, X. Cao, and X. Li, “Rotation invariant point cloud analysis: Where local geometry meets global topology,” in Pattern Recognition, 2022, 127, p. 108626

    2022

  14. [22]

    J. Yu, C. Zhang, and W. Cai, ”Rethinking rotation invariance with point cloud registration,” in AAAI, 2023, vol. 37. No. 3, pp. 3313-3321

    2023

  15. [23]

    The devil is in the pose: Ambiguity-free 3D rotation-invariant learning via pose-aware convolution,

    R. Chen, and C. Yang, “The devil is in the pose: Ambiguity-free 3D rotation-invariant learning via pose-aware convolution,” in CVPR, 2022, pp. 7472–7481

    2022

  16. [24]

    PaRot: Patch-wise rotation- invariant network via feature disentanglement and pose restoration,

    D. Zhang, J. Yu, C. Zhang, and W. Cai, “PaRot: Patch-wise rotation- invariant network via feature disentanglement and pose restoration,” in AAAI, 2023, vol. 37, no. 3, pp. 3418–3426

    2023

  17. [25]

    Rotation-invariant local-to-global repre- sentation learning for 3D point cloud,

    S. Kim, J. Park, and B. Han, “Rotation-invariant local-to-global repre- sentation learning for 3D point cloud,” in NeurIPS, 2020, 33, pp. 8174– 8185

    2020

  18. [26]

    A closer look at rotation-invariant deep point cloud analysis,

    F. Li, K. Fujiwara, F. Okura, and Y . Matsushita, “A closer look at rotation-invariant deep point cloud analysis,” in ICCV, 2021, pp. 16218– 16227

    2021

  19. [27]

    A functional approach to rotation equivariant non-linearities for Tensor Field Networks,

    A. Poulenard, and J.G. Leonidas, “A functional approach to rotation equivariant non-linearities for Tensor Field Networks,” in CVPR, 2021, pp. 13174–13183

    2021

  20. [28]

    SE(3)-Transformers: 3D roto-translation equivariant attention networks,

    F. Fuchs, D. Worrall, V . Fischer, and M. Welling, “SE(3)-Transformers: 3D roto-translation equivariant attention networks,” in NeurIPS, 2022, 33, pp. 1970–1981

    2022

  21. [29]

    A rotation-invariant framework for deep point cloud anal- ysis,

    X. Li et al., “A rotation-invariant framework for deep point cloud anal- ysis,” in IEEE Transactions on Visualization and Computer Graphics , 2021, pp. 4503–4514

    2021

  22. [30]

    Equivariant point cloud analysis via learning orientations for message passing,

    S. Luo, J. Li, J. Guan, Y . Su, C. Cheng, J. Peng, and J. Ma, “Equivariant point cloud analysis via learning orientations for message passing,” in CVPR, 2022, pp. 18932–18941

    2022

  23. [31]

    PointNet: Deep learning on point sets for 3D classification and segmentation,

    C. R. Qi, H. Su, K. Mo, and L. J. Guibas, “PointNet: Deep learning on point sets for 3D classification and segmentation,” in CVPR, 2017, pp. 652–660

    2017

  24. [32]

    PointNet++: Deep hierarchi- cal feature learning on point sets in a metric space,

    C. R. Qi, H. Su, K. Mo, and L. J. Guibas, “PointNet++: Deep hierarchi- cal feature learning on point sets in a metric space,” in NeurIPS, 2017, vol. 30

    2017

  25. [33]

    PointNeXt: Revisiting PointNet++ with improved training and scaling strategies,

    G. Qian, Y . Li, H. Peng, J. Mai, H. Hammoud, M. Elhoseiny and B. Ghanem, “PointNeXt: Revisiting PointNet++ with improved training and scaling strategies,” in NeurIPS, 2022, vol. 35, pp. 23192–23204

    2022

  26. [34]

    PCT: Point cloud transformer,

    M. Guo, J. Cai, Z. Liu, T. Mu, R. Martin and S. Hu, “PCT: Point cloud transformer,” in Computational Visual Media, 2021, vol. 7, pp. 187–199

    2021

  27. [35]

    Walk in the cloud: Learning curves for point clouds shape analysis,

    T. Xiang, C. Zhang, Y . Song, J. Yu and W. Cai, “Walk in the cloud: Learning curves for point clouds shape analysis,” in ICCV, 2021, pp. 915-924

    2021

  28. [36]

    Point transformer,

    H. Zhao, J. Li, J. Jia, H. T. Philip and K. Vladlen, “Point transformer,” in ICCV, 2021, pp. 16259–16268

    2021

  29. [37]

    ShapeNet: An information-rich 3D model repository,

    A.X. Chang, T. Funkhouser, L. Guibas, “ShapeNet: An information-rich 3D model repository,” in arXiv preprint, 2015, arXiv:1512.03012

    2015 arXiv

  30. [38]

    3D ShapeNets: A deep representation for volumetric shapes,

    Z. Wu et al., “3D ShapeNets: A deep representation for volumetric shapes,” in CVPR, 2015, pp. 1912–1920

    2015

  31. [39]

    Revisiting point cloud classification: A new benchmark dataset and classification model on real-world data,

    M. A. Uy, Q. H. Pham, B. S. Hua, T. Nguyen, and S. K. Yeung, “Revisiting point cloud classification: A new benchmark dataset and classification model on real-world data,” in ICCV, 2019, pp. 1588–1597

    2019

  32. [40]

    Self-supervised few-shot learning on point clouds,

    C. Sharma, and M. Kaul, “Self-supervised few-shot learning on point clouds,” in NeurIPS, 2020, pp. 7212–7221

    2020

  33. [41]

    Fixing weight decay regularization in Adam,

    I. Loshchilov, and F. Hutter, “Fixing weight decay regularization in Adam,” in arXiv preprint, 2017, arXiv:1711.05101

    2017 arXiv

  34. [42]

    SGDR: Stochastic gradient descent with warm restarts,

    I. Loshchilov, “SGDR: Stochastic gradient descent with warm restarts,” in arXiv preprint, 2016, arXiv:1608.03983

    2016 arXiv

  35. [43]

    Dynamic graph CNN for learning on point clouds,

    Y . Wang, Y . Sun, Z. Liu, S. E. Sarma, M. M. Bronstein, and J. M. Solomon, “Dynamic graph CNN for learning on point clouds,” in ACM Transactions on Graphics (tog) , 2019, 38(5), pp. 1–12. APPENDIX A. RIHF Details RILF. RILF consists of three types of features that encode inva...

    2019

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.