Pith. sign in

REVIEW 4 major objections 6 minor 53 references

SPAC-Net: Rethinking Point Cloud Completion with Structural Prior

T0 review · 4 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read SPAC-Net claims that locating the boundary between a partial scan and its missing region, then moving those boundary points into the gap, yields state-of-the-art point cloud completion.

desk verdict New interface prior with real efficiency gains, but the ShapeNet SOTA claim is inflated by privileged occlusion-viewpoint information. read the letter →

arxiv 2411.15066 v1 pith:6R5R7EBI submitted 2024-11-22 cs.CV cs.LG

Zizhao Wu , Jian Shi , Xuan Deng , Cheng Zhang , Genfu Yang , Ming Zeng , Yunhai Wang This is my paper · ORCID
classification cs.CVcs.LG
keywords pointcloudcompletionstructuralpriorinterfacedisplacementlearningtransformer3Dshapereconstructiondetailpreservation
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 tries to establish that point cloud completion — inferring a full 3D shape from a partial scan — is done better when the network is guided by an explicit structural prior called the interface: the set of points where the known observation meets the missing region. Instead of generating the coarse shape from a global feature vector, SPAC-Net learns displacements that carry interface points into their corresponding positions in the missing part. A second module, the Structure Supplement (SSP), refines the coarse shape's structural detail before upsampling. On ShapeNet-55/34, PCN, and KITTI, the paper reports lower Chamfer distance and higher F-Score than previous state-of-the-art methods, including PoinTr and SeedFormer. If this is right, explicitly locating the boundary between known and unknown geometry is a stronger inductive bias for completion than learning purely from global shape encodings.

What carries the argument

The load-bearing object is the interface, defined as the set of points lying on the shared boundary between the known partial scan and the missing region; the paper treats the interface as a structural prior because, in the limit of infinitesimal point spacing, the interface computed from the observed side equals the interface on the missing side. The MAD module localizes the interface by viewpoint-distance selection when the occlusion point is known and by a two-threshold edge detector when it is not, and coarse-shape generation is redefined so that each interface point is displaced by a learned MLP-transformed relative feature and added to the interface coordinates. The SSP module then iteratively injects partial-scan detail into the coarse shape via multi-head self- and cross-attention before the FoldingNet upsampler produces the final dense shape.

What would settle it

Run SPAC-Net on ShapeNet-55/34 with the MAD edge detector or an end-to-end interface predictor instead of the known occlusion viewpoint, and compare Chamfer distance and F-Score to the paper's reported numbers; if the margin over SeedFormer and PoinTr disappears, the reported gains depend on privileged viewpoint information, not on the interface prior itself.

Watch

Extended reading notes

Core claim

The paper's central claim is that the boundary between the observed partial scan and the unobserved missing part — the interface — is a reliable structural prior, and that treating coarse-shape generation as a movement of interface points into the missing region preserves local detail that encoder-decoder pipelines lose through max-pooling. The authors define the interface as the intersection between the partial scan and the missing part, argue that both sides share the same boundary as the sampling radius approaches zero, and localize it with a Marginal Detector (MAD) module — using the known occlusion viewpoint on ShapeNet-55/34 and a viewpoint-free edge detector on PCN and KITTI. The predicted coarse shape is then refined by stacked Structure Supplement (SSP) modules that combine self- and cross-attention with the partial-scan features, before a FoldingNet-based upsampler produces the dense completion. The paper reports that SPAC-Net outperforms existing state-of-the-art methods on the three benchmarks, and that adding the interface to a PoinTr baseline improves that baseline as well.

Load-bearing premise

The main benchmark result rests on the assumption that the network knows the exact occlusion viewpoint on ShapeNet-55/34, so it can select interface points by distance to that viewpoint; competing methods do not receive this information, and the paper reports no ShapeNet ablation using its viewpoint-free localization.

Editorial extensions

If this is right

  • If correct, completion networks no longer need to hallucinate missing geometry from a global feature; they can anchor prediction to observed boundary points, which should reduce detail loss in thin structures and edges.
  • Adding the interface as an auxiliary prior to an existing transformer baseline (PoinTr + interface) improves its Chamfer distance and F-Score, suggesting the prior is transferable rather than architecture-specific.
  • The two-phase design — coarse shape via displacement plus SSP refinement before upsampling — implies upsampling modules can concentrate on densification instead of also repairing structure, consistent with the paper's reported lower FLOPs.
  • On unseen categories in ShapeNet-34, the method reports better F-Score than SeedFormer, implying the boundary prior helps generalization to novel object classes.

Reading between the lines

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

  • An implication the paper leaves implicit is that the ShapeNet-55/34 results depend on the occlusion viewpoint being known; the paper does not report a ShapeNet ablation using the viewpoint-free edge detector, so the practical gain in real settings without known viewpoints remains untested.
  • The interface idea could be adapted to other geometry-recovery tasks such as point cloud denoising or inpainting, wherever a boundary between reliable and corrupted regions can be defined.
  • An end-to-end learned interface localizer is a natural next step; the paper notes that its automatic localization experiments underperformed MAD, leaving open whether a better-trained localizer would close the gap to the viewpoint-supervised version.
  • A testable extension would be to add controlled noise to the interface points on synthetic data and measure how completion accuracy degrades, quantifying how much of the gain comes from precise boundary localization.
Share X Bluesky LinkedIn Reddit HN

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 SPAC-Net, a point cloud completion framework built around a new structural prior called the 'interface', defined as the boundary between the observed partial scan and the missing region. A Marginal Detector (MAD) module localizes this interface, either from a known occlusion viewpoint (ShapeNet-55/34) or from a viewpoint-free edge detector (PCN, KITTI). The network then predicts a coarse shape by learning displacements from interface points to their corresponding positions in the missing part, refines the coarse shape with a Structure Supplement (SSP) module, and finally applies a FoldingNet-based upsampler. The paper reports state-of-the-art results on ShapeNet-55/34, PCN, and KITTI, with ablations supporting the interface definition and the number of SSP modules.

Significance. If the results hold, the idea of explicitly conditioning coarse-shape generation on boundary points is a useful and generally applicable contribution to point cloud completion. The PCN and KITTI evaluations use a viewpoint-free interface detector, and the ablations (Tables 6 and 8) support the proposed module choices. The code is publicly available, which aids reproducibility. However, the headline ShapeNet-55/34 comparison is currently compromised by the use of privileged viewpoint information during interface localization, and two core hyperparameters are left unspecified; these issues must be resolved before the claimed state-of-the-art performance can be taken at face value.

major comments (4)
  1. [Sec. 4.3, Tables 1-2] The ShapeNet-55/34 comparison is not apples-to-apples: in these benchmarks the interface is localized by the MAD module using the known occlusion viewpoint that generated each partial scan (Sec. 3.2, case 1), while the baselines (FoldingNet, PCN, PoinTr, SeedFormer) do not receive this privileged information. The paper's own conclusion (Sec. 5) states that automatic end-to-end interface localization 'did not perform as well as the MAD method', which is consistent with the known viewpoint providing a measurable advantage. To support the claim of outperforming state-of-the-art approaches on ShapeNet-55/34, please report results on these benchmarks using the viewpoint-free edge detector of Sec. 3.2 (case 2), or explicitly label the current ShapeNet numbers as oracle-interface results and restrict the headline comparison to the PCN and KITTI benchmarks.
  2. [Sec. 4.6, Table 7] The hyper-parameter δ of the edge detector is selected by evaluating four values on the PCN test set and reporting the one with the best CD-ℓ1 (δ=0.5). This constitutes test-set tuning: the reported 6.61 average is the best of four runs on the test split, which inflates the result relative to the fixed-δ deployment used by the comparison methods. Please select δ on a validation split (or report the results for all four δ values as a sensitivity analysis), and likewise specify the exact split used.
  3. [Sec. 4.2] The number of interface points NT and the edge-detection radius r are core parameters of the MAD module but are never given numeric values. NT is used in Fig. 3 and in the feature-dimension descriptions (NP=NT, CP=256; edgeconv(N=NT, CT=3), etc.), and r is introduced in Sec. 3.2 as the radius controlling the neighborhood, but neither is specified for ShapeNet, PCN, or KITTI. Without these values the experiments cannot be reproduced; please state them explicitly.
  4. [Sec. 3.2, case 1] The localization rule for the occlusion-available scenario states: 'By selecting the nearest NT points in terms of distance from the occlusion point, we can obtain the interface.' Under the benchmark generation described in Sec. 4.1, the partial scan is obtained by removing the N points farthest from a viewpoint, so the interface (the boundary between the partial scan and the missing part) consists of the points in the partial scan with the largest distance from the viewpoint, not the smallest. As written, the rule selects the innermost points rather than the boundary. Please clarify whether the implementation actually selects the farthest points, or explain the geometry in which the nearest points form the boundary.
minor comments (6)
  1. [Abstract] The phrase 'suffer from the loss of details inevitably' is awkward; consider rephrasing for clarity.
  2. [Eq. (4)] The subscript/superscript in FPiTi appears to be a typo; it should be the relative feature FPT between partial scans and interfaces introduced in Sec. 3.2.
  3. [Table 4] The Fidelity values of 0.000 for PoinTr and SPAC-Net are suspicious; please state whether the output always contains the input points, which makes Fidelity trivially zero.
  4. [Sec. 4.5] FLOPs is a count, not a rate; the parenthetical 'Floating Point Operations per Second' is a misnomer and should be corrected.
  5. [Sec. 4.2] The optimizer name is written as 'ADAMW'; it should be 'AdamW'.
  6. [Sec. 3.5] The word 'symetric' should be 'symmetric'.

Circularity Check

0 steps flagged · score 0.0 of 10

No derivation-level circularity: SPAC-Net's residual coarse-shape equation and loss chain are self-contained; the ShapeNet oracle-viewpoint localization and PCN test-set δ selection are evaluation-protocol concerns, not constructional reductions.

full rationale

SPAC-Net's claimed derivation is not circular. The coarse shape is defined as a residual over the interface: Eq. (4) is o_i = γ(β(α(F_{P_i T_i}))) + t_i, with t_i taken from the input partial scan and the network learning an offset toward the missing region; Eq. (7) adds upsampled residuals to that coarse shape, and Eq. (8) supervises both coarse and final outputs against ground truth with Chamfer distance. None of these equations uses the target missing part as an input, and no fitted parameter is obtained by solving for the reported errors. The interface prior is a geometric boundary condition (Secs. 3.1 and 3.2), and its use as an anchor is an inductive bias rather than an identity between input and output. There is no load-bearing self-citation chain and no imported uniqueness theorem. The evaluation protocol does raise legitimate concerns: on ShapeNet-55/34 the interface is localized with the known occlusion viewpoint that generated the partial scans (Secs. 3.2 and 4.3), which is information unavailable to most baselines; the conclusion states automatic end-to-end interface localization underperformed MAD; and δ in Table 7 is tuned on the PCN benchmark, the same CD-ℓ1 metric reported in Table 3. These are fairness, leakage, and reproducibility problems for the SOTA claim, not circular reductions by construction, so on the circularity scale the derivation is self-contained.

Assumptions & free parameters 4 free parameters · 4 assumptions · 1 invented entities

The central claim rests on the geometric assumption of a shared boundary between observed and missing surface, on the availability of a reliable interface localization procedure, and on the expressiveness of the MLP/attention modules used to convert interface displacements into a coarse shape. The model also relies on standard deep learning background (point transformers, EdgeConv, FoldingNet) and on Chamfer distance as an evaluation/training loss.

free parameters (4)
  • edge detection threshold delta = 0.5
    Selected in Sec 4.6/Table 7 by minimizing CD-L1 on the PCN benchmark; influences which points are marked as interface.
  • edge detection radius r = not specified
    Used in the MAD edge detection rule in Sec 3.2 to select neighboring points; the paper gives no value, only says two thresholds r and delta.
  • number of interface points NT = not specified
    Controls the size of the interface set used for displacement prediction; referenced in Fig. 3 and Sec 3.2 but no value is given.
  • number of SSP modules = 3
    Chosen by ablation in Table 8; more modules improve CD up to 3.
assumptions (4)
  • domain assumption The interface of the partial scan is equivalent to the interface of the missing parts, so localizing one gives the other (Sec. 1, Fig. 2b).
    This holds only if the observed and missing regions are complementary closed subsets of the same surface; for discrete point clouds it is an approximation that the paper does not quantify.
  • domain assumption The coarse shape can be expressed as a learned displacement from interface points to missing-part positions using an MLP plus max-pooling (Eq. 4).
    Assumes the network architecture can learn such a mapping from the relative features F_{PiTi}; the paper does not provide a theoretical guarantee, only empirical ablation.
  • domain assumption Chamfer distance between predictions and ground truth is a sufficient training objective and quality measure (Sec. 3.5).
    Standard in the field; Chamfer distance is known to be insensitive to density uniformity in some settings, which the paper partially addresses with F-Score.
  • standard math Multi-head self/cross-attention and transformer layers operate as defined in prior work [35].
    The paper relies on the standard attention mechanism without re-deriving it.
invented entities (1)
  • interface (structural prior)
    purpose: A geometric cue defined as the boundary between observed partial scans and missing regions, used to anchor coarse shape prediction.
    The interface is a modeling construct rather than an independently observed quantity. On ShapeNet it is localized using the ground-truth occlusion viewpoint, and on PCN using a thresholded edge detector; its usefulness is demonstrated only within this paper's pipeline and ablations.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SPAC-Net: Rethinking Point Cloud Completion with Structural Prior." pith.science (2026). https://pith.science/paper/6R5R7EBI

@misc{pith2026241115066,
  author       = {Pith},
  title        = {Pith review of: SPAC-Net: Rethinking Point Cloud Completion with Structural Prior},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6R5R7EBI}},
  note         = {Machine review of arXiv:2411.15066}
}
read the original abstract

Point cloud completion aims to infer a complete shape from its partial observation. Many approaches utilize a pure encoderdecoder paradigm in which complete shape can be directly predicted by shape priors learned from partial scans, however, these methods suffer from the loss of details inevitably due to the feature abstraction issues. In this paper, we propose a novel framework,termed SPAC-Net, that aims to rethink the completion task under the guidance of a new structural prior, we call it interface. Specifically, our method first investigates Marginal Detector (MAD) module to localize the interface, defined as the intersection between the known observation and the missing parts. Based on the interface, our method predicts the coarse shape by learning the displacement from the points in interface move to their corresponding position in missing parts. Furthermore, we devise an additional Structure Supplement(SSP) module before the upsampling stage to enhance the structural details of the coarse shape, enabling the upsampling module to focus more on the upsampling task. Extensive experiments have been conducted on several challenging benchmarks, and the results demonstrate that our method outperforms existing state-of-the-art approaches.

Figures

Figures reproduced from arXiv: 2411.15066 by the authors.

Figure 1
Figure 1. Our method is presented in a comparative visualization with [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. (a) shows several visualization results of interface(yellow points) in incomplete object. (b) illustrates the interface in partial scans is equiv￾alent to the one in missing parts, which means that interface1 equals to interface2, interface3 equals to interface4. So we can localize interface in partial scans to establish spatial perception on missing parts before predicting them. of the coarse shape before upsamplin… view at source ↗
Figure 3
Figure 3. The overall framework of SPAC-Net. Given an input [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Interface Localization Methods Illustration. (a) and (b) depict the [PITH_FULL_IMAGE:figures/full_fig_p003_4.png]
Figure 5
Figure 5. Figure 5: Qualitative comparison of SPAC-Net and state-of-art methods on ShapeNet-55 dataset, showing Airplane, Printer, Chair and Bench from top [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Qualitative results of our method on some novel objects included in ShapeNet-34 dataset. For each object, we provide completion results in [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: The qualitative comparison of our method with other SOTA methods on the PCN dataset [ [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: The qualitative comparison of our method with other SOTA methods on the KITTI dataset [ [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: The visualization of some failure cases. Our method struggles to restore the ideal shape if the input data is severely occluded. [PITH_FULL_IMAGE:figures/full_fig_p009_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 51 canonical work pages

  1. [1]

    Alexiou, E

    E. Alexiou, E. Upenik, and T. Ebrahimi. Towards subjective quality assess- ment of point cloud imaging in augmented reality. In IEEE International Workshop on Multimedia Signal Processing, MMSP, pp. 1–6. IEEE, 2017. 1

    work page 2017

  2. [2]

    Y . Aoki, H. Goforth, R. Srivatsan, and S. Lucey. Pointnetlk: Robust efficient point cloud registration using pointnet. Cornell University - ArXiv, 2019. 1

    work page 2019

  3. [3]

    P. J. Besl and N. D. McKay. A method for registration of 3-d shapes.IEEE Trans. Pattern Anal. Mach. Intell., 14(2):239–256, 1992. 2

    work page 1992

  4. [4]

    Chang, T

    A. Chang, T. Funkhouser, L. Guibas, P. Hanrahan, Q. Huang, Z. Li, S. Savarese, M. Savva, S. Song, H. Su, J. Xiao, L. Yi, and F. Yu. Shapenet: An information-rich 3d model repository. Cornell University - ArXiv ,

    work page

  5. [5]

    X. Chen, K. Jiang, Y . Zhu, X. Wang, and T. Yun. Individual tree crown segmentation directly from uav-borne lidar data using the pointnet of deep learning. Forests, p. 131, Jan 2021. 1

    work page 2021

  6. [6]

    Z. Chen, F. Long, Z. Qiu, T. Yao, W. Zhou, J. Luo, and T. Mei. Anchor- former: Point cloud completion from discriminative nodes. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 13581–13590, 2023. 2

    work page 2023

  7. [7]

    Cheng, M

    X. Cheng, M. Zeng, J. Lin, Z. Wu, and X. Liu. Efficient L0 resampling of point sets. Comput. Aided Geom. Des., 75, 2019. 1

    work page 2019

  8. [8]

    R. Cui, S. Qiu, S. Anwar, J. Liu, C. Xing, J. Zhang, and N. Barnes. P2c: Self-supervised point cloud completion from single partial clouds. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 14351–14360, 2023. 2

    work page 2023

Show all 53 references
  1. [9]

    A. Dai, A. X. Chang, M. Savva, M. Halber, T. A. Funkhouser, and M. Nießner. Scannet: Richly-annotated 3d reconstructions of indoor scenes. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR, pp. 2432–2443. IEEE Computer Society, 2017. 1

    2017

  2. [10]

    A. Dai, C. R. Qi, and M. Nießner. Shape completion using 3d-encoder- predictor cnns and shape synthesis. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR, pp. 6545–6554. IEEE Computer Society, 2017. 1, 2

    2017

  3. [11]

    L. Fu, J. Liu, J. Zhou, M. Zhang, and Y . Lin. Tree skeletonization for raw point cloud exploiting cylindrical shape prior. IEEE Access , 8:27327–27341, Feb 2020. 3

    2020

  4. [12]

    Geiger, P

    A. Geiger, P. Lenz, C. Stiller, and R. Urtasun. Vision meets robotics: The KITTI dataset. Int. J. Robotics Res., 32(11):1231–1237, 2013. 1, 2, 3, 5, 8

    2013

  5. [13]

    Groueix, M

    T. Groueix, M. Fisher, V . G. Kim, B. C. Russell, and M. Aubry. Atlasnet: a papier-mâché approach to learning 3d surface generation (2018). ArXiv preprint ArXiv:1802.05384, 11, 1802. 5

    2018 arXiv

  6. [14]

    Guo, J.-X

    M.-H. Guo, J.-X. Cai, Z.-N. Liu, T.-J. Mu, R. R. Martin, and S.-M. Hu. Pct: Point cloud transformer. Computational Visual Media, p. 187–199, Apr 2021. 2

    2021

  7. [15]

    X. Han, Z. Li, H. Huang, E. Kalogerakis, and Y . Yu. High-resolution shape completion using deep neural networks for global structure and local geometry inference. In IEEE International Conference on Computer Vision, ICCV, pp. 85–93. IEEE Computer Society, 2017. 1, 2

    2017

  8. [16]

    Huang, Y

    Z. Huang, Y . Yu, J. Xu, F. Ni, and X. Le. Pf-net: Point fractal network for 3d point cloud completion. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR, pp. 7659–7667. Computer Vision Foundation / IEEE, 2020. 1, 2, 5

    2020

  9. [17]

    Krenzin and O

    J. Krenzin and O. Hellwich. Reduction of point cloud artifacts using shape priors estimated with the gaussian process latent variable model. In Pattern Recognition: 38th German Conference, GCPR 2016, Hannover, Germany, September 12-15, 2016, Proceedings 38, pp. 273–284. Springer,

    2016

  10. [18]

    S. Li, P. Gao, X. Tan, and M. Wei. Proxyformer: Proxy alignment assisted point cloud completion with missing part sensitive transformer. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 9466–9475, 2023. 1, 2

    2023

  11. [19]

    X. Li, L. Wang, and Y . Fang. Unsupervised category-specific partial point set registration via joint shape completion and registration. IEEE Trans. Vis. Comput. Graph., 29(7):3251–3265, 2023. 1

    2023

  12. [20]

    Z. Liu, H. Tang, Y . Lin, and S. Han. Point-voxel CNN for efficient 3d deep learning. In Advances in Neural Information Processing Systems, pp. 963–973, 2019. 1, 2

    2019

  13. [21]

    Loshchilov

    I. Loshchilov. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101, 2017. 5

    2017 arXiv

  14. [22]

    Mirikharaji and G

    Z. Mirikharaji and G. Hamarneh. Star shape prior in fully convolutional networks for skin lesion segmentation. In Medical Image Computing and Computer Assisted Intervention–MICCAI, pp. 737–745, 2018. 3

    2018

  15. [23]

    D. T. Nguyen, B. Hua, M. Tran, Q. Pham, and S. Yeung. A field model for repairing 3d shapes. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR, pp. 5676–5684. IEEE Computer Society, 2016. 1, 2

    2016

  16. [24]

    C. R. Qi, H. Su, K. Mo, and L. J. Guibas. Pointnet: Deep learning on point sets for 3d classification and segmentation. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR, pp. 77–85. IEEE Computer Society, 2017. 1, 2

    2017

  17. [25]

    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, pp. 5099–5108, 2017. 1, 2, 4

    2017

  18. [26]

    G. Qian, Y . Li, H. Peng, J. Mai, H. Hammoud, M. Elhoseiny, and B. Ghanem. Pointnext: Revisiting pointnet++ with improved training and scaling strategies. Advances in Neural Information Processing Sys- tems, 35:23192–23204, 2022. 1

    2022

  19. [27]

    Y . Rao, Y . Nie, and A. Dai. Patchcomplete: Learning multi-resolution patch priors for 3d shape completion on unseen categories. Advances in Neural Information Processing Systems, 35:34436–34450, 2022. 3

    2022

  20. [28]

    Sharma, O

    A. Sharma, O. Grau, and M. Fritz. Vconv-dae: Deep volumetric shape learning without object labels. In Computer Vision–ECCV 2016 Work- shops, pp. 236–250. Springer, 2016. 1, 2

    2016

  21. [29]

    Z. Shi, Z. Chen, Z. Xu, W. Yang, Z. Yu, and L. Huang. Shape prior guided attack: Sparser perturbations on 3d point clouds. In Proceedings of the AAAI Conference on Artificial Intelligence, pp. 8277–8285, 2022. 3

    2022

  22. [30]

    Stutz and A

    D. Stutz and A. Geiger. Learning 3d shape completion under weak super- vision. Int. J. Comput. Vis., 128(5):1162–1181, 2020. 1, 2

    2020

  23. [31]

    Tatarchenko, S

    M. Tatarchenko, S. R. Richter, R. Ranftl, Z. Li, V . Koltun, and T. Brox. What do single-view 3d reconstruction networks learn? In IEEE Confer- ence on Computer Vision and Pattern Recognition, CVPR, pp. 3405–3414,

  24. [32]

    L. P. Tchapmi, V . Kosaraju, H. Rezatofighi, I. D. Reid, and S. Savarese. Topnet: Structural point cloud decoder. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR, pp. 383–392. Computer Vision Foundation / IEEE, 2019. 2, 5

    2019

  25. [33]

    M. Tian, M. H. Ang, and G. H. Lee. Shape prior deformation for cate- gorical 6d object pose and size estimation. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XXI 16, pp. 530–546, 2020. 3

    2020

  26. [34]

    Varley, C

    J. Varley, C. DeChant, A. Richardson, J. Ruales, and P. K. Allen. Shape completion enabled robotic grasping. In IEEE/RSJ International Con- ference on Intelligent Robots and Systems, IROS, pp. 2442–2447. IEEE,

  27. [35]

    Vaswani, N

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser, and I. Polosukhin. Attention is all you need. In Advances in Neural Information Processing Systems, pp. 5998–6008, 2017. 1, 2

    2017

  28. [36]

    X. Wang, M. H. Ang, and G. H. Lee. Cascaded refinement network for point cloud completion. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR, pp. 787–796, 2020. 5

    2020

  29. [37]

    X. Wang, M. H. Ang, and G. H. Lee. Point cloud completion by learning shape priors. In 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 10719–10726. IEEE, 2020. 3

    2020

  30. [38]

    Y . Wang, Y . Sun, Z. Liu, S. E. Sarma, M. M. Bronstein, and J. M. Solomon. Dynamic graph CNN for learning on point clouds. ACM Trans. Graph., 38(5):146:1–146:12, 2019. 1, 4, 5

    2019

  31. [39]

    Y . Wang, D. J. Tan, N. Navab, and F. Tombari. Learning local displace- ments for point cloud completion. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR, pp. 1568–1577, 2022. 2

    2022

  32. [40]

    X. Wen, P. Xiang, Z. Han, Y . Cao, P. Wan, W. Zheng, and Y . Liu. Pmp-net: Point cloud completion by learning multi-step point moving paths. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR, pp. 7443–7452, 2021. 5

    2021

  33. [41]

    Xiang, X

    P. Xiang, X. Wen, Y . Liu, Y . Cao, P. Wan, W. Zheng, and Z. Han. Snowflak- enet: Point cloud completion by snowflake point deconvolution with skip- transformer. In IEEE/CVF International Conference on Computer Vision, ICCV, pp. 5479–5489, 2021. 5

    2021

  34. [42]

    H. Xie, H. Yao, S. Zhou, J. Mao, S. Zhang, and W. Sun. Grnet: Grid- ding residual network for dense point cloud completion. In European Conference on Computer Vision, pp. 365–381. Springer, 2020. 2, 4, 5, 7

    2020

  35. [43]

    Y . Yang, C. Feng, Y . Shen, and D. Tian. Foldingnet: Point cloud auto- encoder via deep grid deformation. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR, pp. 206–215, 2018. 2, 4, 5, 6

    2018

  36. [44]

    X. Yu, Y . Rao, Z. Wang, Z. Liu, J. Lu, and J. Zhou. Pointr: Diverse point cloud completion with geometry-aware transformers. In IEEE/CVF International Conference on Computer Vision, ICCV, pp. 12478–12487. IEEE, 2021. 1, 2, 4, 5, 6, 7

    2021

  37. [45]

    W. Yuan, T. Khot, D. Held, C. Mertz, and M. Hebert. PCN: point com- pletion network. In International Conference on 3D Vision, 3DV , pp. 728–737. IEEE Computer Society, 2018. 2, 4, 5, 6, 7, 8

    2018

  38. [46]

    Zhang, X

    B. Zhang, X. Zhao, H. Wang, and R. Hu. Shape completion with points in the shadow. In S. K. Jung, J. Lee, and A. W. Bargteil, eds., SIGGRAPH Asia, pp. 28:1–28:9. ACM, 2022. 3

    2022

  39. [47]

    Zhang, H

    C. Zhang, H. Chen, H. Wan, P. Yang, and Z. Wu. Graph-pbn: Graph-based parallel branch network for efficient point cloud learning. Graph. Model., 119:101120, 2022. 1

    2022

  40. [48]

    Zhang, Q

    W. Zhang, Q. Yan, and C. Xiao. Detail preserved point cloud completion via separated feature aggregation. In Computer Vision - ECCV 2020 - 16th European Conference, Glasgow, UK, August 23-28, 2020, Proceedings, Part XXV, vol. 12370 of Lecture Notes in Computer Science, pp. 512...

    2020

  41. [49]

    Zhang, H

    W. Zhang, H. Zhou, Z. Dong, J. Liu, Q. Yan, and C. Xiao. Point cloud completion via skeleton-detail transformer. IEEE Trans. Vis. Comput. Graph., 29(10):4229–4242, 2023. 2

    2023

  42. [50]

    H. Zhao, L. Jiang, J. Jia, P. H. Torr, and V . Koltun. Point transformer. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 16259–16268, 2021. 4

    2021

  43. [51]

    X. Zhao, B. Zhang, J. Wu, R. Hu, and T. Komura. Relationship-based point cloud completion. IEEE Trans. Vis. Comput. Graph., 28(12):4940–4950,

  44. [52]

    Zheng, Y

    Y . Zheng, Y . Li, S. Yang, and H. Lu. Global-pbnet: A novel point cloud registration for autonomous driving. IEEE Trans. Intell. Transp. Syst. , 23(11):22312–22319, 2022. 1

    2022

  45. [53]

    H. Zhou, Y . Cao, W. Chu, J. Zhu, T. Lu, Y . Tai, and C. Wang. Seedformer: Patch seeds based point cloud completion with upsample transformer. In Computer Vision - ECCV, vol. 13663 of Lecture Notes in Computer Science, pp. 416–432, 2022. 1, 2, 4, 5, 6, 7

    2022

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.