Pith. sign in

REVIEW 4 major objections 7 minor 31 references

SEDD-PCC: A Single Encoder-Dual Decoder Framework For End-To-End Learned Point Cloud Compression

T0 review · 4 major / 7 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read A single encoder can map a point cloud's color and geometry into one shared latent space, and two decoders reconstruct both, beating G-PCC by 75.0% in geometry bitrate.

desk verdict The architecture is genuinely new, but the reported bitrate savings may not include the G-PCC-coded thumbnail geometry side channel, so the headline numbers are not yet established. read the letter →

arxiv 2505.16709 v1 pith:VZTHNSGD submitted 2025-05-22 cs.CV eess.IV

Kai Hsiang Hsieh , Monyneath Yim , Jui Chiu Chiang This is my paper · ORCID
classification cs.CVeess.IV
keywords learnedpointcloudcompressionjointgeometryandattributecodingsingleencoderdualdecodersharedlatentspaceknowledgedistillationsparseconvolutionBD-BRPCQM
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 is trying to establish that point cloud geometry and attributes can be compressed jointly by one encoder and two decoders, rather than by separate codecs. It claims a three-channel, attribute-only input still leaves a shared latent that supports accurate geometry reconstruction, eliminating the recoloring step used by sequential pipelines. The headline numbers are BD-BR savings of 75.0% for D1-PSNR, 32.6% for Y-PSNR, and 33.2% for 1-PCQM versus G-PCC on MPEG test sequences, with a model of 32.6 MB. A sympathetic reader cares because this is a simpler, lighter architecture than sequential joint codecs, and it goes end-to-end.

What carries the argument

The core mechanism is the shared latent space $z$ obtained by an attribute-oriented sparse-convolution encoder built with Inception-Residual layers. The encoder sees a sparse voxel tensor with coordinates $C$ and RGB features $F$, so the latent jointly represents appearance and structure. A transform module adapts $z$ into geometry-specific features for the geometry decoder, which uses a Top-k+1 classification mechanism to reconstruct point positions, while the attribute decoder follows the Sparse-PCAC process. The three-stage training protocol, attribute coding, geometry coding with teacher-model knowledge distillation, and joint fine-tuning, resolves the conflicting objectives, and the loss is a Lagrangian rate-distortion objective with lambda weighting.

What would settle it

Train SEDD-PCC on point clouds whose RGB values are replaced by a constant while keeping the same coordinates, and compare D1-PSNR on the 8iVFB test set. If geometry quality is maintained, the geometric information is coming from the sparse coordinates rather than from the attribute-derived latent, undercutting the claim that the shared latent carries geometry; if it collapses, the attribute-centric encoder is genuinely load-bearing.

Watch

Extended reading notes

Core claim

SEDD-PCC establishes that a single sparse-convolution encoder, fed only the three color channels of a voxelized point cloud, can produce a shared latent representation from which a transform module plus a geometry decoder reconstructs point positions while an attribute decoder reconstructs colors. The paper claims this is the first single-encoder, dual-decoder architecture for joint geometry and attribute compression, and that it avoids the recoloring stage of sequential pipelines. Compared with G-PCC, it reports average BD-BR reductions of 75.0% for D1-PSNR, 32.6% for Y-PSNR, and 33.2% for 1-PCQM, while staying competitive with learned sequential codecs.

Load-bearing premise

The load-bearing premise is that the three attribute channels, after passing through one shared encoder, leave a latent representation rich enough for the transform module and geometry decoder to reconstruct point positions accurately; if attribute-oriented features wash out geometric detail, the reported D1-PSNR gains collapse.

Editorial extensions

If this is right

  • SEDD-PCC removes the separate recoloring stage, so attribute quality no longer inherits geometry reconstruction errors through a recoloring step.
  • A single attribute-centric encoder suffices for both tasks, so joint codecs can be built with fewer parameters than two separate codecs; the model is 32.6 MB.
  • Knowledge distillation from a geometry teacher and a dedicated transform module each provide roughly 7% bitrate savings, making them effective components for shared-latent joint compression.
  • The design offers an end-to-end learned alternative to G-PCC with an average 75.0% D1-PSNR bitrate reduction and 33.2% 1-PCQM reduction on the evaluated test sequences.
  • Because geometry and attributes share one latent, rate-distortion optimization can be done jointly in a single loss, avoiding separate bit allocation between two codecs, though the paper notes this allocation may not yet be optimal.

Reading between the lines

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

  • The paper leaves implicit that the same single-encoder design could serve point clouds without color by setting attributes to a constant; a natural test is whether geometry-only BD-BR still beats G-PCC under that input.
  • Because bit allocation is set by the loss weights instead of a rate-allocation algorithm, the reported average BD-BR may shift along the geometry/attribute trade-off frontier; comparing SEDD-PCC against an optimal allocation frontier would show how much headroom remains.
  • The claimed independence from recoloring implies that geometry errors no longer leak into attribute quality; a direct test is to compare Y-PSNR of SEDD-PCC against a sequential codec at equal geometry distortion.
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 / 7 minor

Summary. The paper introduces SEDD-PCC, a unified learned point cloud codec with a single encoder and two decoders for geometry and attributes. Instead of representing the input as a four-channel occupancy-plus-color voxel grid, the encoder takes only the three RGB attribute channels, and a transform module plus a geometry decoder reconstruct the point positions from the shared latent. Training is done in three stages: attribute coding, geometry coding with knowledge distillation from PCGCv2, and joint fine-tuning. On 8iVFB and Owlii sequences, the method reports BD-BR savings of 75.0% (D1-PSNR), 32.6% (Y-PSNR), and 33.2% (1-PCQM) relative to G-PCC v23, with a 32.6 MB model and lower complexity than sequential approaches.

Significance. If the reported numbers are sound, SEDD-PCC offers a compelling demonstration that a single attribute-centric encoder can support both geometry and attribute decoding in a lightweight, end-to-end learned pipeline, avoiding the recoloring overhead of sequential codecs. The three-stage training and the use of a transform module with knowledge distillation are sensible engineering choices, and the ablation study (Table 3) provides initial evidence for their contribution. However, the manuscript currently omits critical experimental details—most importantly the bitrate accounting for the G-PCC-coded thumbnail geometry C_z—and several comparisons are incomplete, so the central claim of large BD-BR savings is not yet established. The paper also lacks absolute rate-distortion points and error bars, which are standard expectations in learned compression papers.

major comments (4)
  1. [Section 2.2, Eqs. (1)-(3), Table 1] The encoding process losslessly encodes a thumbnail point cloud geometry C_z with the G-PCC octree codec, yet the rate R in all training losses is defined as R_zhat, the rate of the quantized feature F_z. The paper never states whether the bitrate of C_z is included in the reported BD-BR numbers of Table 1. If it is excluded, the claimed 75.0% bitrate reduction for D1-PSNR over G-PCC is not a fair end-to-end comparison, since G-PCC's bitrate includes full geometry. If it is included, the paper must report the bitrate split between C_z and F_z, otherwise the reader cannot determine whether the geometry-quality gains come from the learned geometry decoder or from the G-PCC side information. Because the central contribution is that a single attribute-only encoder extracts shared features for geometry reconstruction, this missing accounting is load-bearing.
  2. [Section 3.1, Table 1] Table 1 reports only BD-BR values and no absolute rate-distortion operating points. Without the underlying RD points, the BD-BR computations cannot be verified, and the reader cannot assess whether the average savings are dominated by a particular rate region. The authors should provide the per-sequence RD data (e.g., for D1-PSNR, Y-PSNR, 1-PCQM) in supplementary material.
  3. [Table 1, JPEG Pleno column] In Table 1, the JPEG Pleno columns contain dashes for five of the six test sequences, with numbers only for Soldier, yet the final row reports an 'Average' of -62.9, -51.5, -42.3 for D1, Y, 1-PCQM. This average appears to be computed from the single Soldier row, not from an average over sequences. Reporting a single-sequence result as an 'Average' is misleading and should be corrected; the comparison to JPEG Pleno is otherwise incomplete. Similarly, the Unicorn method is mentioned in Section 3.2 but no results for it appear in Table 1 or Fig. 4, so the claim of 'competitive performance against other learning-based methods' is not fully supported.
  4. [Section 2.1, Section 2.2] The paper claims that using only three attribute channels in the shared encoder is sufficient because 'attributes inherently encapsulate richer information than the geometry occupancy,' but it provides no evidence that the latent representation retains the geometric structure needed for accurate D1-PSNR reconstruction. Since the geometry decoder operates on the latent F_z, and the paper does not isolate the contribution of F_z versus the G-PCC-coded C_z, the role of the shared latent in geometry reconstruction is unclear. An ablation that removes or corrupts C_z (or compares with a pure G-PCC geometry side stream) would directly address this concern.
minor comments (7)
  1. [Section 3.1] The hyperparameter lists for λ_A, λ_G, λ_t in stage 3 are given as six values but it is unclear how many rate points are generated and whether they correspond to the six columns of Fig. 4; please clarify the mapping between λ values and rate points.
  2. [Eq. (2)] The term ℒ_BCE2 is described as 'multi-scale BCE' but the multi-scale formulation is not specified; please define it consistently with the main BCE loss.
  3. [Table 2] The encoding/decoding times are missing for several methods with only dashes shown; please either provide the values or state that they were not measured.
  4. [Section 2.2] The generation of the thumbnail geometry C_z is not described (e.g., voxel size, whether it is derived from the input coordinates or from the latent); please provide this detail.
  5. [References] Reference [5] contains a typo 'J, -J. Chiu' and should be 'J.-C. Chiu'.
  6. [Fig. 4] The axes and units of the rate-distortion plot are not legible in the text; please ensure the figure is high resolution with labeled axes.
  7. [Section 3.4.2 and Fig. 5] The text states that without the transform module 'the bitrate is reduced by 7.1%' but Table 3 reports -7.1 for D1, which indicates a bitrate saving; the wording is ambiguous, as it could be misinterpreted as an increase; please rephrase to 'bitrate is reduced (improved) by 7.1%'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central compression gains are measured against external G-PCC and PCGCv2 anchors, and no load-bearing prediction reduces to its own inputs by construction.

full rationale

SEDD-PCC is an empirical systems paper. The headline BD-BR numbers in Table 1 are computed on held-out 8iVFB and Owlii test sequences against the external G-PCC TMC13 v23 anchor, and the Stage-2 teacher model is the independently published PCGCv2 [3]. The loss functions in Eqs. (1)-(3) are ordinary R+lambda-D Lagrangian objectives; none of the reported metrics is obtained by fitting a parameter to the target and then renaming the fit as a prediction. The two self-citations ([5], [11]) appear only in the related-work survey and are not load-bearing for the architecture, training, or ablation conclusions. One non-circular validity concern: Section 2.2 sends a G-PCC-octree-coded thumbnail geometry C_z as side information while the rate terms R_zhat in Eqs. (1)-(3) omit the C_z bitrate, and Table 3 does not ablate C_z. This is an accounting/completeness risk for the absolute BD-BR claim, not a case where an output is equivalent to an input by definition. Under the hard rule that circularity requires an exhibited Eq.-to-Eq. reduction or a fitted-input-renamed-as-prediction, this gap does not raise the circularity score.

Assumptions & free parameters 8 free parameters · 5 assumptions · 0 invented entities

The central performance claim rests on a standard Lagrangian rate-distortion training objective, on the assumption that an attribute-only encoder retains enough geometry for the dual decoders, on the choice of many weighting hyperparameters, and on comparability of benchmark numbers from external papers. None of these are formal axioms; they are engineering assumptions that should be stated explicitly.

free parameters (8)
  • voxelization_bit_depth = 6-bit per dimension
    Input point clouds are partitioned into non-overlapping 6-bit-sized cubes (Section 3.1); this discretization bounds the finest recoverable detail and is a manual choice.
  • alpha = 2
    Weights the two MSE terms in the attribute loss Eq. (1); chosen by hand.
  • lambda_A_stage1 = 0.03
    Lagrangian weight for attribute distortion in Eq. (1); chosen by hand.
  • lambda_G_stage2 = 6
    Lagrangian weight for geometry distortion in Eq. (2); chosen by hand.
  • lambda_MSE_stage2 = 1.5
    Weight for feature distillation loss in Eq. (2); chosen by hand.
  • lambda_A_stage3 = 0.03, 0.04, 0.04, 0.05, 0.05, 0.05
    Per-rate attribute weights in the joint fine-tuning loss Eq. (3); tuned to produce six rate points.
  • lambda_G_stage3 = 6, 4, 4, 8, 12, 20
    Per-rate geometry weights in Eq. (3); tuned to produce six rate points.
  • lambda_t_stage3 = 0.5, 0.25, 0.125, 0.05, 0.015, 0.005
    Overall rate-distortion trade-off in Eq. (3); tuned to produce six rate points.
assumptions (5)
  • domain assumption Sparse voxel grid at 6-bit resolution captures all geometry and attributes needed for the test benchmarks.
    Section 3.1: all inputs are voxelized into 6-bit cubes before coding; any detail below this resolution cannot be reconstructed, so the reported D1/Y/PCQM numbers are conditional on this discretization.
  • domain assumption The shared encoder, fed only three attribute channels, produces latent features from which geometry can be decoded.
    Section 2.1: the encoder processes only RGB channels; the geometry decoder and transform module assume that these attribute features retain sufficient geometric structure. The paper does not quantify how much geometry information survives in the latent space.
  • domain assumption Knowledge distillation from the PCGCv2 teacher is a valid and non-circular training signal.
    Section 2.3, Eq. (2): student features are matched to teacher features; this assumes the teacher feature space is compatible with the shared encoder's latent space.
  • domain assumption BD-BR results from YOGA, DeepPCC, and JPEG Pleno are directly comparable with SEDD-PCC under the same anchor.
    Section 3.2, Table 1: numbers are taken from external papers; no re-evaluation under a common test condition is reported, and some table entries are missing.
  • domain assumption The G-PCC octree coding of the thumbnail geometry C_z is a fixed, non-learned component that does not dominate the comparison unfairly.
    Section 2.2: thumbnail geometry is losslessly coded with G-PCC; the method is not fully end-to-end learned for geometry coordinates. The comparison with G-PCC as anchor still includes this standard codec internally.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SEDD-PCC: A Single Encoder-Dual Decoder Framework For End-To-End Learned Point Cloud Compression." pith.science (2026). https://pith.science/paper/VZTHNSGD

@misc{pith2026250516709,
  author       = {Pith},
  title        = {Pith review of: SEDD-PCC: A Single Encoder-Dual Decoder Framework For End-To-End Learned Point Cloud Compression},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VZTHNSGD}},
  note         = {Machine review of arXiv:2505.16709}
}
read the original abstract

To encode point clouds containing both geometry and attributes, most learning-based compression schemes treat geometry and attribute coding separately, employing distinct encoders and decoders. This not only increases computational complexity but also fails to fully exploit shared features between geometry and attributes. To address this limitation, we propose SEDD-PCC, an end-to-end learning-based framework for lossy point cloud compression that jointly compresses geometry and attributes. SEDD-PCC employs a single encoder to extract shared geometric and attribute features into a unified latent space, followed by dual specialized decoders that sequentially reconstruct geometry and attributes. Additionally, we incorporate knowledge distillation to enhance feature representation learning from a teacher model, further improving coding efficiency. With its simple yet effective design, SEDD-PCC provides an efficient and practical solution for point cloud compression. Comparative evaluations against both rule-based and learning-based methods demonstrate its competitive performance, highlighting SEDD-PCC as a promising AI-driven compression approach.

Figures

Figures reproduced from arXiv: 2505.16709 by the authors.

Figure 5
Figure 5. The network of the alternative geometry decoder [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

31 extracted references · 29 canonical work pages

  1. [1]

    V-PCC codec description, in ISO/IEC JTC 1/SC 29/WG 7 N00100, 2021

    work page 2021

  2. [2]

    G-PCC codec description , in ISO/IEC JTC 1/SC 29/WG 7 N00271, 2022

    work page 2022

  3. [3]

    Multiscale point cloud geometry compression,

    J. Wang, D. Ding, Z. Li and Z. Ma, “Multiscale point cloud geometry compression,” Proc. of Data Compression Conference (DCC), 2021, pp. 73-82

    work page 2021

  4. [4]

    Sparse tensor-based multiscale representation for point cloud geometry compression,

    J. Wang, D. Ding, Z. Li, X. Feng, C. Cao and Z. Ma, “Sparse tensor-based multiscale representation for point cloud geometry compression,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 45, no. 7, pp. 9055 - 9071, 1 July 2023

    work page 2023

  5. [5]

    ANFPCGC++: Point cloud geometry coding using augmented normalizing flows and Transformer-based entropy model,

    J. -C. Chiang, J, -J. Chiu, and M. Yim, “ANFPCGC++: Point cloud geometry coding using augmented normalizing flows and Transformer-based entropy model,” IEEE Access, 2024

    work page 2024

  6. [6]

    GRASP-Net: Geometric residual analysis and synthesis for point cloud compression,

    J. Pang, M. A. Lodhi, and D. Tian, “GRASP-Net: Geometric residual analysis and synthesis for point cloud compression,” ACM MM, 2022

    work page 2022

  7. [7]

    PCGFormer: Lossy point cloud geometry compression via local self -attention,

    G. Liu, J. Wang, D. Ding and Z. Ma, “PCGFormer: Lossy point cloud geometry compression via local self -attention,” Proc. of IEEE International Conference on Visual Communications and Image Processing (VCIP), 2022

    work page 2022

  8. [8]

    Deep - PCAC: An end-to-end deep lossy compression framework for point cloud attributes,

    X. Sheng, L. Li, D. Liu, Z. Xiong, Z. Li and F. Wu, “Deep - PCAC: An end-to-end deep lossy compression framework for point cloud attributes,” IEEE Transactions on Mul timedia, vol. 24, pp. 2617-2632, 2022

    work page 2022

Show all 31 references
  1. [9]

    Sparse tensor -based point cloud attribute compression ,

    J. Wang and Z. Ma, “Sparse tensor -based point cloud attribute compression ,” Proc. of IEEE International Conference on Multimedia Information Processing and Retrieval, 2022, pp. 59-64

    2022

  2. [10]

    Reducing the complexity of normalizing flow architectures for point cloud attribute compression ,

    R. B. Pinheiro, J. -E. Marvie, G. Valenzise and F. Dufaux, “Reducing the complexity of normalizing flow architectures for point cloud attribute compression ,” Proc. of IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2024

    2024

  3. [11]

    Sparse tensor -based point cloud attribute compression using augmented normalizing flows,

    T. -P. Lin, M. Yim, J. -C. Chiang, W. -H. Peng and W. -N. Lie, “Sparse tensor -based point cloud attribute compression using augmented normalizing flows,” Proc. of Asia Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC), 2023, pp. 1...

    2023

  4. [12]

    SPAC: Sampling -based Progressive Attribute Compression for Dense Point Clouds ,

    X. Mao, H . Yuan, T. Guo, S . Jiang, R . Hamzaoui and S. Kwong, “SPAC: Sampling -based Progressive Attribute Compression for Dense Point Clouds ,” arXiv preprint arXiv:2409.10293, 2024

    2024 arXiv

  5. [13]

    ResNeRF-PCAC: Super Resolving Residual Learning NeRF for High Efficiency Point Cloud Attributes Coding ,

    S. Umair, B. Kathariya, Z. Li, A. Akhtar , G. Van der Auwera, “ResNeRF-PCAC: Super Resolving Residual Learning NeRF for High Efficiency Point Cloud Attributes Coding ,” Proc. of IEEE International Conference on Image Processing (ICIP), 2024, pp. 3540-3546

    2024

  6. [14]

    Towards neural network approaches for point cloud compression,

    E. Alexiou, K. Tung, and T. Ebrahimi, “Towards neural network approaches for point cloud compression,” Proc. of Applications of digital image processing XLIII , vol. 11510, pp. 18-37, 2020

    2020

  7. [15]

    Deep Learning-Based Point Cloud Coding and Super -Resolution: a Joint Geometry and Color approach,

    A. F. R. Guarda, M. Ruivo, L. Coelho, A. Seleem, N. M. M. Rodrigues and F. Pereira, “Deep Learning-Based Point Cloud Coding and Super -Resolution: a Joint Geometry and Color approach,” IEEE Trans. Multimedia , doi: 10.1109/TMM.2023.3338081

  8. [16]

    YOGA: Yet another geometry-based point clou d compressor,

    J. Zhang, T. Chen, D. Ding, and Z. Ma, “YOGA: Yet another geometry-based point clou d compressor,” Proc. of ACM International Conference on Multimedia , 2023, pp. 9070 - 9081

    2023

  9. [17]

    DeepPCC: Learned Lossy Point Cloud Compression,

    J. Zhang, G. Liu, J. Zhang, D. Ding, and Z. Ma, “DeepPCC: Learned Lossy Point Cloud Compression, ” accepted by IEEE Trans. Emerging Topics in Computational Intelligence, Sept. 2024

    2024

  10. [18]

    Verification Model Description for JPEG Pleno Learning - based Point Cloud Coding v4.0,

    “Verification Model Description for JPEG Pleno Learning - based Point Cloud Coding v4.0,” in ISO/IEC JTC1/SC29/WG1 N100709, 102nd Meeting, San Francisco, CA, USA, Jan. 2024

    2024

  11. [19]

    The JPEG Pleno Learning-based Point Cloud Coding Standard: Serving Man and Machine,

    A. F. Guarda, N. M. Rodrigues, and F. Pereira, “The JPEG Pleno Learning-based Point Cloud Coding Standard: Serving Man and Machine,” arXiv preprint arXiv:2409.08130, 2024

    2024

  12. [20]

    A Versatile Point Cloud Compressor Using Universal Multiscale Conditional Coding - Part I: Geometry,

    J. Wang, R . Xue, J . Li, D . Ding, Y. Lin, and Z . Ma, “A Versatile Point Cloud Compressor Using Universal Multiscale Conditional Coding - Part I: Geometry, ” IEEE Trans. Pattern Analysis and Machine Intelligence, 47(1):269- 287, Jan. 2025

    2025

  13. [21]

    A Versatile Point Cloud Compressor Using Universal Multiscale Conditional Coding - Part II: Attribute,

    J. Wang, R . Xue, J . Li, D . Ding, Y. Lin, and Z . Ma, “A Versatile Point Cloud Compressor Using Universal Multiscale Conditional Coding - Part II: Attribute, ” IEEE Trans. Pattern Analysis and Machine Intelligence, 47(1):252- 268, Jan. 2025

    2025

  14. [22]

    The JPEG AI standard: providing efficient human and machine visual data consumption,

    J. Ascenso, E. Alshina and T. Ebrahimi, “The JPEG AI standard: providing efficient human and machine visual data consumption,” IEEE MultiMedia, vol. 30, no. 1, pp. 100-111, 1 Jan.-Mar. 2023, doi: 10.1109/MMUL.2023.3245919

    2023

  15. [23]

    Inception-v4, inception -resnet and the impact of residual connections on learning,

    C. Szegedy, S. Ioffe, V. Vanhoucke and A. Alemi, "Inception-v4, inception -resnet and the impact of residual connections on learning," In Proceedings of the AA AI conference on artificial intelligence (Vol. 31, No. 1)

  16. [24]

    SparsePCGCv1 Update: Improvements on dense/sparse/LiDAR point clouds,

    “SparsePCGCv1 Update: Improvements on dense/sparse/LiDAR point clouds, ” ISO/IEC JTC 1/SC 29/WG 7 m60352, Online, July. 2022

    2022

  17. [25]

    ScanNet: richly -annotated 3d reconstructions of indoor scenes,

    A. Dai, A. X. Chang, M. Savva, M. Halber, T. Funkhouser and M. Nießner, “ScanNet: richly -annotated 3d reconstructions of indoor scenes,” IEEE Conference on Computer Vision and Pattern Recognition , 2017, pp. 2432 - 2443

    2017

  18. [26]

    4D spatio-temporal ConvNets: Minkowski convolutional neural networks,

    C. Choy, J. Gwak and S. Savarese, “4D spatio-temporal ConvNets: Minkowski convolutional neural networks,” Proc. of IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019, pp. 3070-3079

    2019

  19. [27]

    8I voxelized full bodies - a voxelized point cloud dataset,

    E. d’Eon, B. Harrison, T. Myers, and P. A. Chou, “8I voxelized full bodies - a voxelized point cloud dataset,” ISO/IEC JTC1/SC29 Joint WG11/WG1 (MPEG/JPEG) m40059/M74006, Jan. 2017

    2017

  20. [28]

    X. Yi, L. Yao, and W. Ziyu, “ Owlii Dynamic Human Mesh Sequence Dataset, ISO/IEC Standard JTC1/SC29/WG11 m41658, ISO/IEC: Washington, DC, USA, Oct. 2017

    2017

  21. [29]

    PCQM: A full-reference quality metric for colored 3D point clouds,

    G. Meynet, Ya. Nehmé, J. Digne, and G. Lavoué. “PCQM: A full-reference quality metric for colored 3D point clouds,” Proc. of International Conference on Quality of Multimedia Experience (QoMEX), 2020

    2020

  22. [30]

    Quality Analysis of the Coding Bitrate Tradeoff Between Geometry and Attributes for Colored Point Clouds

    J. Prazeres, R. Rodrigues, M. Pereira and A. M. G. Pinheiro, “Quality Analysis of the Coding Bitrate Tradeoff Between Geometry and Attributes for Colored Point Clouds. ” arXiv preprint arXiv:2410.21613, 2024

    2024 arXiv

  23. [31]

    Common test conditions for point cloud compression,

    “Common test conditions for point cloud compression,” in ISO/IEC JTC1/SC29/WG11 MPEG output document N19084

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.