Pith. sign in

REVIEW 4 major objections 6 minor 55 references

Point2SpatialCapsule: Aggregating Features and Spatial Relationships of Local Regions on Point Clouds using Spatial-aware Capsules

T0 review · 4 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read Point2SpatialCapsule reports 93.7% on ModelNet40 by replacing max-pooling with routing between spatial-aware capsules.

desk verdict A plausible incremental architecture that combines NetVLAD-style clustering with capsule routing, but the paper's SOTA and robustness claims outrun its evidence — the motivating invariance to rigid transforms is contradicted by Eq. (2) as written. read the letter →

arxiv 1908.11026 v1 pith:SUTHU4MZ submitted 2019-08-29 cs.CV

Xin Wen , Zhizhong Han , Xinhai Liu , Yu-Shen Liu This is my paper · ORCID
classification cs.CV
keywords pointclouds3Dshaperepresentationcapsulenetworksdynamicroutingfeatureaggregationspatialrelationshipsclassificationpartsegmentation
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

The paper argues that the standard way of turning a point cloud into a single vector—extracting features from local regions and then max-pooling them—throws away the spatial relationships between regions, and that this loss limits recognition accuracy. It proposes Point2SpatialCapsule, which first clusters local-region features into learnable centers and clusters the point coordinates separately, then concatenates the two to form feature-spatial embeddings. These embeddings are turned into spatial-aware capsules, and a dynamic-routing layer aggregates them into a global representation. The paper reports that this representation outperforms pooling-based networks and prior capsule networks on ModelNet40 classification (93.4% with xyz, 93.7% with normals), ModelNet10 classification and retrieval, and ShapeNet part segmentation. The point is that routing between clusters that carry explicit spatial coordinates can preserve arrangement information that max-pooling filters out.

What carries the argument

The load-bearing object is the spatial-aware capsule: a capsule built from a feature-spatial embedding formed by concatenating a soft-assigned feature cluster center with a soft-assigned coordinate cluster center. The embedding is rearranged and squashed into short vectors, and dynamic routing (the coupling-coefficient update procedure from capsule networks) aggregates them into digit capsules, the class-level output vectors. The geometric feature aggregation module's clustering is what binds the routing log priors to fixed locations in feature space, so routing can encode spatial relationships despite the disorder of raw points.

What would settle it

Train on the aligned ModelNet40 split, then evaluate on a version of the test set rotated by arbitrary 3D rotations with no retraining; if accuracy falls far below the max-pooling baseline instead of staying near 93.4%, the spatial relationships the routing learns are tied to absolute coordinates rather than intrinsic part arrangements.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that dynamic routing in a capsule network can be made to learn spatial relationships between local regions of a point cloud, provided the capsules are built from cluster centers that encode absolute 3D locations, rather than from a single pooled global feature. The paper claims that previous capsule applications to point clouds lose spatial information because they build capsules after max-pooling or fully-connected aggregation. Point2SpatialCapsule instead performs soft-assignment clustering of both local features and point coordinates, concatenates the resulting embeddings, splits them into short vectors combined with the spatial embedding, and runs dynamic routing. The paper reports state-of-the-art results on 3D shape classification, retrieval, and segmentation on the ModelNet and ShapeNet datasets, and takes this as evidence that the spatial-aware capsules capture arrangement information that pooling-based methods filter out.

Load-bearing premise

The network assumes each point cloud has been aligned and normalized in advance, so absolute 3D coordinates are meaningful; a rotated or shifted cloud changes the coordinate clusters and the routing weights that the whole argument depends on.

Editorial extensions

If this is right

  • Any pooling-based point-cloud network should be able to swap its final max-pooling for the spatial-aware capsule aggregator and gain discriminative power, since the aggregator is designed as a drop-in replacement after local feature extraction.
  • Retrieval becomes a distance computation between capsule length vectors, which the paper reports reaches 89.43% mAP on ModelNet40.
  • On ShapeNet part segmentation, the arrangement-aware global feature reaches 85.3% mean IoU, matching or exceeding several local-feature methods even though the capsule module was not designed for per-point features.
  • The spatial relationship aggregation is most effective when paired with the geometric feature aggregation: removing the clustering module is the largest single ablation loss in the paper, dropping accuracy from 93.4% to 91.4%.

Reading between the lines

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

  • A direct test of the paper's core claim would be to perturb poses: because the geometric feature aggregation clusters raw coordinates, arbitrary rotations should displace the spatial embeddings; if accuracy on rotated test clouds collapses, the learned spatial relationships are pose-relative rather than intrinsic arrangements.
  • If the arrangement information is really carried by the cluster centers and log priors, a simpler architecture—learned positional encodings plus self-attention over the same clusters—might reproduce the gains without dynamic routing; comparing the two would isolate what the capsule routing specifically adds.
  • The same feature-spatial clustering pattern could transfer to other set-structured inputs with meaningful absolute positions, such as 2D keypoint sets or unordered graphs with node coordinates, where pooling is also the default aggregator.
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

4 major / 6 minor

Summary. The paper proposes Point2SpatialCapsule, a point-cloud representation learning architecture that replaces max-pooling aggregation of local region features with two modules: geometric feature aggregation, which uses NetVLAD-style soft assignment to cluster both local features and raw 3D coordinates into learnable centers, and spatial relationship aggregation, which applies dynamic routing between spatial-aware capsules built from the concatenated feature-spatial embeddings. The representation is trained with a margin loss and a chamfer reconstruction loss, and is evaluated on ModelNet40/10 classification, ModelNet retrieval, and ShapeNet part segmentation. The paper claims state-of-the-art performance on classification, retrieval, and segmentation tasks.

Significance. If the claims were fully supported, the contribution would be significant: it provides a concrete way to make dynamic routing usable for unordered point sets by anchoring capsules to learned cluster centers, and it addresses a real limitation of max-pooling aggregation. The paper is generally clear and the architecture is well motivated, with ablations for each component, and the use of standard benchmarks is appropriate. However, the empirical evidence as presented does not establish the headline claim: the best plain-xyz ModelNet40 result is below RS-CNN, segmentation results are below RS-CNN and PointCNN, and no uncertainty estimates or pose-robustness tests are reported. The central idea is plausible, but the current validation is insufficient for the claimed scope.

major comments (4)
  1. [Abstract, Sec. IV-B, Table I] The headline claim that Point2SpatialCapsule 'outperforms the state-of-the-art methods' is not supported by Table I: with 1024x3 input, the method obtains 93.4% on ModelNet40, while RS-CNN reports 93.6%, and the paper's own text concedes that the method is 'ranked the second place.' The additional claim that RS-CNN uses ten voting tests is not sufficient to establish superiority, because no corresponding controlled comparison (e.g., with test-time augmentation disabled) is provided. At minimum, the claims should be scoped to the settings where the method is actually best (ModelNet10, and ModelNet40 with normals), and the comparison protocol should be documented.
  2. [Sec. III-B.2, Eq. (2), Fig. 2] The design motivation in Fig. 2 states that the clustering addresses 'the shifting and rotation of point cloud,' but Eq. (2) constructs spatial embeddings C(y_k) from absolute coordinates x_i and fixed learned centers y_k. Under a rigid transform of the input, every residual x_i - y_k changes, so the spatial embeddings, the feature-spatial embeddings, and the routing log priors all change. The paper describes no canonicalization step and reports no experiment with rotated or translated test clouds in Sec. IV; therefore the claimed 'relatively invariant' behavior of the geometric feature aggregation is not demonstrated by the architecture as written. The authors should either add a pose-robustness experiment (e.g., ModelNet40 with random rotation/translation at test time) or remove/qualify the invariance motivation in Fig. 2.
  3. [Sec. IV-E, Tables IV-VII] The ablation study is reported without error bars or multiple-run statistics, and the differences are small: Table IV gives 92.5, 91.4, 92.1, and 93.44 for the four variants, and Table V gives 93.44, 92.22, and 91.98 for routing iterations. Differences of 0.5 to 1.3 percentage points are within typical run-to-run variation for point-cloud networks, so the conclusion that 'each part contributes to the model performance' is not statistically supported. Please report mean and standard deviation over at least three runs, or a paired significance test, for the main ablation and comparison tables.
  4. [Sec. IV-D, Table III] The segmentation results also contradict the overview claim: the mean instance IoU is 85.3, below RS-CNN (86.2) and PointCNN (86.1). The text acknowledges this only indirectly by saying the method produces 'comparable results,' but the abstract and introduction claim superiority over state-of-the-art in segmentation. The claims should be revised to state that the method is competitive rather than superior on this task.
minor comments (6)
  1. [Secs. IV-B, IV-C, IV-D] Typographical errors: 'Point2Capusule' appears several times; the intended name is Point2SpatialCapsule.
  2. [Sec. III-B.2] The definition C(s_k) = [C(y_k) : C(x_k)] uses C(x_k), but the feature embedding was defined as C(q_k); please make the notation consistent.
  3. [Sec. III-B.2, Eq. (2)] The soft-assignment weights in Eq. (2) use the same w_k as Eq. (1) with a different bias b'_k; please clarify whether the weights are shared and define b'_k.
  4. [Sec. IV-A.1] The paper reports a ModelNet10 split of 2,468 training and 909 testing samples; the standard split used in the cited references is 3,991 training and 908 testing, so please verify and correct this sentence.
  5. [Sec. IV-E, Table IV and Table I] Table IV reports Full-Model accuracy as 93.44 while Table I reports 93.4; please unify the precision of reported numbers.
  6. [Sec. IV-D] The sentence 'the experimental results prove that Point2Sequence improves the quality of local feature extraction' appears to mean Point2SpatialCapsule; please correct the name.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's empirical claims are tested on held-out ModelNet/ShapeNet benchmarks against external baselines, and its self-citations are contextual rather than load-bearing.

full rationale

The paper's central claims are architectural and empirical: Point2SpatialCapsule aggregates local features and coordinates into learnable cluster centers (Eqs. 1 and 2), then applies dynamic routing between the resulting spatial-aware capsules (Eqs. 4-7). No equation defines a target result in terms of itself, and no fitted parameter is renamed as a prediction. Classification, retrieval, and segmentation accuracies are measured on standard held-out test sets with external reference numbers, so the 'state-of-the-art' claim is falsifiable and not circular. The ablations (No-Multi, No-VLAD, No-Caps) compare architecture variants rather than fitting a parameter and then 'predicting' a closely related quantity. The authors cite several of their own prior works, but these are contextual related-work references (e.g., Point2Sequence, SeqViews2SeqLabels) and are not used as evidence for the main contribution; the load-bearing capsule routing and clustering machinery cite Sabour et al. and NetVLAD, which are external. The strongest concern raised in the skeptic note—that the spatial embeddings in Eq. (2) depend on absolute coordinates and may be sensitive to rotation or translation—is a robustness/validity risk, not a circularity, because the experiments use canonical aligned inputs and the method's performance is measured directly rather than derived from its assumptions. Overall, the derivation chain is self-contained against external benchmarks: the paper's equations and experiments do not presuppose the conclusions they are used to support.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The paper does not introduce physical entities or data-fitting parameters in the sense of material science. The listed free parameters are network hyperparameters selected by hand or by ablation. The key assumptions are the canonical alignment of input clouds and the equating of routing log priors with spatial relationships, both of which are design premises rather than independently verified facts.

free parameters (5)
  • Number of cluster centers Q = 64
    Number of feature and coordinate cluster centers in the geometric feature aggregation module; chosen by ablation over {16, 32, 64, 128} (Table VI), not derived from theory.
  • Multi-scale shuffling ratio r = 2
    Ratio used to reshape the T x C feature tensor into rT x (C/r); set to 2 in Sec. IV-A.3 with no ablation study reported for this value.
  • Reconstruction loss weight alpha = 0.0001
    Weight of the chamfer reconstruction loss in the total loss; selected after ablation over {1e-3, 1e-4, 1e-5, 0} (Table VII).
  • Dynamic routing iterations = 1
    Number of dynamic routing iterations; chosen after ablation over {1, 3, 5}, with 1 iteration giving the best accuracy (Table V).
  • Capsule configuration = 1024 capsules of 16 dimensions
    Each feature-spatial embedding is split into 16 16-dimensional vectors, producing 1024 spatial-aware capsules; this is a hand-set architectural choice in Sec. IV-A.3.
assumptions (3)
  • domain assumption Input point clouds are pre-aligned and normalized to a canonical coordinate frame.
    The method clusters absolute 3D coordinates in Sec. III-B.2 to form spatial embeddings. If the cloud is rotated or translated, the coordinates and hence the log priors change. The paper does not test robustness to arbitrary rotations, and ModelNet data are typically provided in a canonical orientation.
  • ad hoc to paper Dynamic routing log priors learned over feature cluster centers encode spatial relationships among local regions.
    This is the central premise stated in Sec. III-C.2 and Fig. 2: because features are bound to fixed cluster centers, routing learns log priors that reflect spatial configuration. The paper argues this qualitatively but provides no direct measurement or ablation isolating this effect from the extra model capacity of the routing layers.
  • domain assumption Reported baseline results from prior publications are accurate and obtained under settings comparable to the proposed method.
    The comparison tables (Tables I-III) list numbers from other papers without re-running those methods under identical conditions. The paper acknowledges one discrepancy (RS-CNN uses ten voting tests) but otherwise assumes the quoted figures are reliable.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Point2SpatialCapsule: Aggregating Features and Spatial Relationships of Local Regions on Point Clouds using Spatial-aware Capsules." pith.science (2026). https://pith.science/paper/SUTHU4MZ

@misc{pith2026190811026,
  author       = {Pith},
  title        = {Pith review of: Point2SpatialCapsule: Aggregating Features and Spatial Relationships of Local Regions on Point Clouds using Spatial-aware Capsules},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SUTHU4MZ}},
  note         = {Machine review of arXiv:1908.11026}
}
read the original abstract

Learning discriminative shape representation directly on point clouds is still challenging in 3D shape analysis and understanding. Recent studies usually involve three steps: first splitting a point cloud into some local regions, then extracting corresponding feature of each local region, and finally aggregating all individual local region features into a global feature as shape representation using simple max pooling. However, such pooling-based feature aggregation methods do not adequately take the spatial relationships between local regions into account, which greatly limits the ability to learn discriminative shape representation. To address this issue, we propose a novel deep learning network, named Point2SpatialCapsule, for aggregating features and spatial relationships of local regions on point clouds, which aims to learn more discriminative shape representation. Compared with traditional max-pooling based feature aggregation networks, Point2SpatialCapsule can explicitly learn not only geometric features of local regions but also spatial relationships among them. It consists of two modules. To resolve the disorder problem of local regions, the first module, named geometric feature aggregation, is designed to aggregate the local region features into the learnable cluster centers, which explicitly encodes the spatial locations from the original 3D space. The second module, named spatial relationship aggregation, is proposed for further aggregating clustered features and the spatial relationships among them in the feature space using the spatial-aware capsules developed in this paper. Compared to the previous capsule network based methods, the feature routing on the spatial-aware capsules can learn more discriminative spatial relationships among local regions for point clouds, which establishes a direct mapping between log priors and the spatial locations through feature clusters.

Figures

Figures reproduced from arXiv: 1908.11026 by the authors.

Figure 1
Figure 1. The illustration of comparison between the max-pooling based [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The illustration of directly applying capsule network to the point [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. The architecture of our proposed Point2SpatialCapsule. For input point clouds, (a) the multi-scale local feature extraction first extracts features from [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Illustration of feature similarity between adjacent points. Features of [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Illustration of the multi-scale shuffling, which is the solution to the [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Comparison of the strategies for applying dynamic routing in local region features between (a) Point2SpatialCapsule and (b) the previous methods [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Illustration of rearrange and squashing layer. The green block is the [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Illustration of the segmentation network in our Point2SpatialCapsule. [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: The comparison of precision and recall curves obtained by different [PITH_FULL_IMAGE:figures/full_fig_p009_9.png]
Figure 10
Figure 10. Figure 10: Visualization of part segmentation results. In each shape pair, the first row is the ground truth (GT), and the second row is our predicted result. Parts [PITH_FULL_IMAGE:figures/full_fig_p010_10.png]
Figure 11
Figure 11. Figure 11: The visualization of reconstruction results on the test set of ModelNet40. The top roll is the input original point cloud, and the bottom roll is the [PITH_FULL_IMAGE:figures/full_fig_p011_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 52 canonical work pages

  1. [1]

    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 Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2017

    work page 2017

  2. [2]

    3DViewGraph: Learning global features for 3D shapes from a graph of unordered views with attention,

    Z. Han, X. Wang, C.-M. V ong, Y .-S. Liu, M. Zwicker, and C. Chen, “3DViewGraph: Learning global features for 3D shapes from a graph of unordered views with attention,” in International Joint Conference on Artificial Intelligence, 2019

    work page 2019

  3. [3]

    A-CNN: Annularly convolu- tional neural networks on point clouds,

    A. Komarichev, Z. Zhong, and J. Hua, “A-CNN: Annularly convolu- tional neural networks on point clouds,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2019, pp. 7421–7430

    work page 2019

  4. [4]

    View inter-prediction GAN: Unsupervised representation learning for 3D shapes by learning global shape memories to support local view predictions,

    Z. Han, M. Shang, Y .-S. Liu, and M. Zwicker, “View inter-prediction GAN: Unsupervised representation learning for 3D shapes by learning global shape memories to support local view predictions,” in 33rd AAAI Conference on Artificial Intelligence , 2019

    work page 2019

  5. [5]

    Parts4Feature: Learning 3D global features from generally semantic parts in multiple views,

    Z. Han, X. Liu, Y .-S. Liu, and M. Zwicker, “Parts4Feature: Learning 3D global features from generally semantic parts in multiple views,” in International Joint Conference on Artificial Intelligence , 2019

    work page 2019

  6. [6]

    SeqViews2SeqLabels: Learning 3D global features via aggregating sequential views by RNN with attention,

    Z. Han, M. Shang, Z. Liu, C.-M. V ong, Y .-S. Liu, J. Han, M. Zwicker, and C. P. Chen, “SeqViews2SeqLabels: Learning 3D global features via aggregating sequential views by RNN with attention,” IEEE Transac- tions on Image Processing , vol. 28, no. 2, pp. 658–672, 2019

    work page 2019

  7. [7]

    Point2Sequence: Learning the shape representation of 3D point clouds with an attention-based sequence to sequence network,

    X. Liu, Z. Han, Y .-S. Liu, and M. Zwicker, “Point2Sequence: Learning the shape representation of 3D point clouds with an attention-based sequence to sequence network,” in 33rd AAAI Conference on Artificial Intelligence, 2019

    work page 2019

  8. [8]

    Octree guided CNN with spherical kernels for 3D point clouds,

    H. Lei, N. Akhtar, and A. Mian, “Octree guided CNN with spherical kernels for 3D point clouds,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2019

    work page 2019

Show all 55 references
  1. [9]

    Associatively segmenting instances and semantics in point clouds,

    X. Wang, S. Liu, X. Shen, C. Shen, and J. Jia, “Associatively segmenting instances and semantics in point clouds,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2019, pp. 4096–4105

    2019

  2. [10]

    3D-SIS: 3D semantic instance segmentation of RGB-D scans,

    J. Hou, A. Dai, and M. Nießner, “3D-SIS: 3D semantic instance segmentation of RGB-D scans,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2019, pp. 4421–4430

    2019

  3. [11]

    Escape from cells: Deep kd-networks for the recognition of 3D point cloud models,

    R. Klokov and V . Lempitsky, “Escape from cells: Deep kd-networks for the recognition of 3D point cloud models,” in IEEE International Conference on Computer Vision (ICCV) , 2017, pp. 863–872

    2017

  4. [12]

    PointNet++: Deep hierarchical feature learning on point sets in a metric space,

    C. R. Qi, L. Yi, H. Su, and L. J. Guibas, “PointNet++: Deep hierarchical feature learning on point sets in a metric space,” in Advances in Neural Information Processing Systems , 2017, pp. 5099–5108

    2017

  5. [13]

    PointCNN: Con- volution on x-transformed points,

    Y . Li, R. Bu, M. Sun, W. Wu, X. Di, and B. Chen, “PointCNN: Con- volution on x-transformed points,” in Advances in Neural Information Processing Systems, 2018, pp. 820–830

    2018

  6. [14]

    SO-Net: Self-organizing network for point cloud analysis,

    J. Li, B. M. Chen, and G. H. Lee, “SO-Net: Self-organizing network for point cloud analysis,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2018, pp. 9397–9406

    2018

  7. [15]

    Mining point cloud local structures by kernel correlation and graph pooling,

    Y . Shen, C. Feng, Y . Yang, and D. Tian, “Mining point cloud local structures by kernel correlation and graph pooling,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2018

    2018

  8. [16]

    Dynamic routing between capsules,

    S. Sabour, N. Frosst, and G. E. Hinton, “Dynamic routing between capsules,” in Advances in Neural Information Processing Systems, 2017, pp. 3856–3866

    2017

  9. [17]

    SpiderCNN: Deep learning on point sets with parameterized convolutional filters,

    Y . Xu, T. Fan, M. Xu, L. Zeng, and Y . Qiao, “SpiderCNN: Deep learning on point sets with parameterized convolutional filters,” in Proceedings of the European Conference on Computer Vision (ECCV) , 2018, pp. 87–102

    2018

  10. [18]

    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,” arXiv:1801.07829, 2018

    2018 arXiv

  11. [19]

    Relation-shape convolutional neural network for point cloud analysis,

    Y . Liu, B. Fan, S. Xiang, and C. Pan, “Relation-shape convolutional neural network for point cloud analysis,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2019, pp. 8895–8904

    2019

  12. [20]

    PointWeb: Enhancing local neighborhood features for point cloud processing,

    H. Zhao, L. Jiang, C.-W. Fu, and J. Jia, “PointWeb: Enhancing local neighborhood features for point cloud processing,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2019, pp. 5565–5573

    2019

  13. [21]

    O-CNN: Octree-based convolutional neural networks for 3D shape analysis,

    P.-S. Wang, Y . Liu, Y .-X. Guo, C.-Y . Sun, and X. Tong, “O-CNN: Octree-based convolutional neural networks for 3D shape analysis,” ACM Transactions on Graphics , vol. 36, no. 4, p. 72, 2017

    2017

  14. [22]

    OctNet: Learning deep 3D representations at high resolutions,

    G. Riegler, A. O. Ulusoy, and A. Geiger, “OctNet: Learning deep 3D representations at high resolutions,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2017

    2017

  15. [23]

    PVNet: A joint convolutional network of point cloud and multi-view for 3D shape Recognition,

    H. You, Y . Feng, R. Ji, and Y . Gao, “PVNet: A joint convolutional network of point cloud and multi-view for 3D shape Recognition,” in Proceedings of the 26th ACM international conference on Multimedia , 2018, pp. 1310–1318

    2018

  16. [24]

    3D2SeqViews: Aggregating sequential views for 3D global feature learning by CNN with hierarchical attention aggregation,

    Z. Han, H. Lu, Z. Liu, C.-M. V ong, Y .-S. Liu, M. Zwicker, J. Han, and C. P. Chen, “3D2SeqViews: Aggregating sequential views for 3D global feature learning by CNN with hierarchical attention aggregation,” IEEE Transactions on Image Processing, vol. 28, no. 8, pp. 3986–3999, 2019

    2019

  17. [25]

    GIFT: Towards scalable 3D shape retrieval,

    S. Bai, X. Bai, Z. Zhou, Z. Zhang, Q. Tian, and L. J. Latecki, “GIFT: Towards scalable 3D shape retrieval,”IEEE Transactions on Multimedia, vol. 19, no. 6, pp. 1257–1271, 2017

    2017

  18. [26]

    Multi-Angle Point cloud- V AE: Unsupervised feature learning for 3D point clouds from multiple angles by joint self-reconstruction and half-to-half prediction,

    Z. Han, X. Wang, Y .-S. Liu, and M. Zwicker, “Multi-Angle Point cloud- V AE: Unsupervised feature learning for 3D point clouds from multiple angles by joint self-reconstruction and half-to-half prediction,” in IEEE International Conference on Computer Vision (ICCV) , 2019

    2019

  19. [27]

    Y2Seq2Seq: Cross-modal representation learning for 3D shape and text by joint reconstruction and prediction of view and word sequences,

    Z. Han, M. Shang, X. Wang, Y .-S. Liu, and M. Zwicker, “Y2Seq2Seq: Cross-modal representation learning for 3D shape and text by joint reconstruction and prediction of view and word sequences,” The Thirty- Third AAAI Conference on Artificial Intelligence (AAAI) , 2019

    2019

  20. [28]

    DeeppPano: Deep panoramic representation for 3-d shape recognition,

    B. Shi, S. Bai, Z. Zhou, and X. Bai, “DeeppPano: Deep panoramic representation for 3-d shape recognition,” IEEE Signal Processing Letters, vol. 22, no. 12, pp. 2339–2343, 2015

    2015

  21. [29]

    Exploiting the PANORAMA representation for convolutional neural network classification and retrieval,

    S. K, T. T, and I. Pratikakis, “Exploiting the PANORAMA representation for convolutional neural network classification and retrieval,” in 3DOR, 2017

    2017

  22. [30]

    Computing the inner distances of volumetric models for articulated shape description with a visibility graph,

    Y .-S. Liu, K. Ramani, and M. Liu, “Computing the inner distances of volumetric models for articulated shape description with a visibility graph,” IEEE Transactions on Pattern Analysis and Machine Intelli- gence, vol. 23, no. 12, pp. 2538–2544, 2011

    2011

  23. [31]

    Robust shape normalization of 3D articulated volumetric models,

    C. Wang, Y .-S. Liu, M. Liu, J.-H. Yong, and J.-C. Paul., “Robust shape normalization of 3D articulated volumetric models,” Computer-Aided Design, vol. 44, no. 12, pp. 1253–1268, 2012

    2012

  24. [32]

    3D ShapeNets: A deep representation for volumetric shapes,

    Z. Wu, S. Song, A. Khosla, F. Yu, L. Zhang, X. Tang, and J. Xiao, “3D ShapeNets: A deep representation for volumetric shapes,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2015, pp. 1912–1920

    2015

  25. [33]

    Unsupervised learning of 3D local features from raw voxels based on a novel permutation voxelization strategy,

    Z. Han, Z. Liu, J. Han, C. V ong, S. Bu, and C. Chen, “Unsupervised learning of 3D local features from raw voxels based on a novel permutation voxelization strategy,” IEEE Transactions on Cybernetics , vol. 49, no. 2, pp. 481–494, 2019. JOURNAL OF LATEX CLASS FILES, VOL. 14, N...

    2019

  26. [34]

    VConv-DAE: Deep volumetric shape learning without object labels,

    A. Sharma, O. Grau, and M. Fritz, “VConv-DAE: Deep volumetric shape learning without object labels,” in European Conference on Computer Vision. Springer, 2016, pp. 236–250

    2016

  27. [35]

    DeepShape: Deep learned shape descriptor for 3D shape matching and retrieval,

    J. Xie, Y . Fang, F. Zhu, and E. Wong, “DeepShape: Deep learned shape descriptor for 3D shape matching and retrieval,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2015, pp. 1275–1283

    2015

  28. [36]

    Learned binary spectral shape descriptor for 3D shape correspondence,

    J. Xie, M. Wang, and Y . Fang, “Learned binary spectral shape descriptor for 3D shape correspondence,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2016, pp. 3309–3317

    2016

  29. [37]

    BoSCC: Bag of spatial context correlations for spatially enhanced 3D shape representation,

    Z. Han, Z. Liu, C.-M. V ong, Y .-S. Liu, S. Bu, J. Han, and C. P. Chen, “BoSCC: Bag of spatial context correlations for spatially enhanced 3D shape representation,” IEEE Transactions on Image Processing, vol. 26, no. 8, pp. 3707–3720, 2017

    2017

  30. [38]

    Deep Spatiality: Unsupervised learning of spatially-enhanced global and local 3D features by deep neural network with coupled softmax,

    ——, “Deep Spatiality: Unsupervised learning of spatially-enhanced global and local 3D features by deep neural network with coupled softmax,” IEEE Transactions on Image Processing , vol. 27, no. 6, pp. 3049–3063, 2018

    2018

  31. [39]

    CapsuleGAN: Generative adversarial capsule network,

    A. Jaiswal, W. Abdalmageed, Y . Wu, and P. Natarajan, “CapsuleGAN: Generative adversarial capsule network,” inProceedings of the European Conference on Computer Vision , 2018

    2018

  32. [40]

    Capsules for object segmentation,

    R. Lalonde and U. Bagci, “Capsules for object segmentation,” arXiv:1804.04241, 2018

    2018 arXiv

  33. [41]

    Investigating capsule networks with dynamic routing for text classification,

    M. Yang, W. Zhao, J. Ye, Z. Lei, Z. Zhao, and S. Zhang, “Investigating capsule networks with dynamic routing for text classification,” in Pro- ceedings of the Conference on Empirical Methods in Natural Language Processing, 2018, pp. 3110–3119

    2018

  34. [42]

    MCapsNet: Capsule network for text with multi-task learning,

    L. Xiao, H. Zhang, W. Chen, Y . Wang, and Y . Jin, “MCapsNet: Capsule network for text with multi-task learning,” in Proceedings of the Conference on Empirical Methods in Natural Language Processing , 2018, pp. 4565–4574

    2018

  35. [43]

    Attention-based capsule networks with dynamic routing for relation extraction,

    Z. Ningyu, D. Shumin, S. Zhanlin, C. Xi, Z. Wei, and C. Huajun, “Attention-based capsule networks with dynamic routing for relation extraction,” in Proceedings of the Conference on Empirical Methods in Natural Language Processing, 2018

    2018

  36. [44]

    Object classification from 3D volumetric data with 3D capsule networks,

    K. Burak, A. Ayesha, and V . Senem, “Object classification from 3D volumetric data with 3D capsule networks,” in IEEE Global Conference on Signal and Information Processing , 2018

    2018

  37. [45]

    3D Point Capsule Networks,

    Y . Zhao, T. Birdal, H. Deng, and F. Tombari, “3D Point Capsule Networks,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2019

    2019

  38. [46]

    3DCapsule: Extending the capsule architecture to classify 3D point clouds,

    C. Ali and P. Lars, “3DCapsule: Extending the capsule architecture to classify 3D point clouds,” in IEEE Winter Conference on Applications of Computer Vision , 2019

    2019

  39. [47]

    PointNetVLAD: Deep point cloud based retrieval for large-scale place recognition,

    M. A. Uy and G. H. Lee, “PointNetVLAD: Deep point cloud based retrieval for large-scale place recognition,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2018, pp. 4470–4479

    2018

  40. [48]

    NetVLAD: CNN architecture for weakly supervised place recognition,

    R. Arandjelovic, P. Gronat, A. Torii, T. Pajdla, and J. Sivic, “NetVLAD: CNN architecture for weakly supervised place recognition,” in Pro- ceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, pp. 5297–5307

    2016

  41. [49]

    Attentional ShapeContextNet for point cloud recognition,

    S. Xie, S. Liu, Z. Chen, and Z. Tu, “Attentional ShapeContextNet for point cloud recognition,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2018, pp. 4606–4615

    2018

  42. [50]

    Real-time single image and video super- resolution using an efficient sub-pixel convolutional neural network,

    W. Shi, J. Caballero, F. Husz ´ar, J. Totz, A. P. Aitken, R. Bishop, D. Rueckert, and Z. Wang, “Real-time single image and video super- resolution using an efficient sub-pixel convolutional neural network,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Re...

    2016

  43. [51]

    SHREC’16 track large-scale 3D shape retrieval from ShapeNet core55,

    M. Savva, F. Yu, H. Su, M. Aono, B. Chen, D. Cohen-Or, W. Deng, H. Su, S. Bai, X. Bai et al. , “SHREC’16 track large-scale 3D shape retrieval from ShapeNet core55,” in Proceedings of the Eurographics Workshop on 3D Object Retrieval , 2016

    2016

  44. [52]

    Modeling point clouds with self-attention and gumbel subset sampling,

    J. Yang, Q. Zhang, B. Ni, L. Li, J. Liu, M. Zhou, and Q. Tian, “Modeling point clouds with self-attention and gumbel subset sampling,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2019, pp. 3323–3332

    2019

  45. [53]

    PointConv: Deep convolutional networks on 3D point clouds,

    W. Wu, Z. Qi, and L. Fuxin, “PointConv: Deep convolutional networks on 3D point clouds,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2019, pp. 9621–9630

    2019

  46. [54]

    V olumetric and multi-view CNNs for object classification on 3D data,

    C. R. Qi, H. Su, M. Niebner, A. Dai, M. Yan, and L. J. Guibas, “V olumetric and multi-view CNNs for object classification on 3D data,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, pp. 5648–5656

    2016

  47. [55]

    SPNet: Deep 3D object clas- sification and retrieval using stereographic projection,

    M. Yavartanoo, E. Y . Kim, and K. M. Lee, “SPNet: Deep 3D object clas- sification and retrieval using stereographic projection,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2018. Xin Wen received the B.S. degree in engineering management f...

    2018

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.