Pith. sign in

REVIEW 4 major objections 5 minor 55 references

ResPCC: A Loss-Resilient Neural Point Cloud Codec over Lossy Networks

T0 review · 4 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read ResPCC claims that point cloud codecs can be made intrinsically resilient to packet loss by conditioning on loss rate and restoring scattered missing latent features; tests on ShapeNet and SemanticKITTI at 5–30% loss support this.

desk verdict Solid loss-resilient point cloud codec with a real, fixable reporting flaw; the burst-loss sensitivity gap is the main thing to probe before believing the strong practical claims. read the letter →

arxiv 2608.11845 v1 pith:PV6HJ5SV submitted 2026-08-12 eess.IV cs.MM

Xueqin Niu , Mufan Liu , Yifan Wang , Le Yang , Jun Sun , Yiling Xu This is my paper · ORCID
classification eess.IVcs.MM
keywords pointcloudcompressionlossresiliencepacketlatentrestorationerrorconcealmentrate-distortionoptimizationinterleavingneuralcodec
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

ResPCC is a learned point cloud codec that treats packet loss as part of the coding problem rather than an afterthought. The paper claims it is the first end-to-end neural codec with intrinsic loss resilience for intra-frame geometry: a single model, conditioned on an estimated loss rate, keeps reconstruction quality high when 5% to 30% of packets are lost. The design scatters loss damage across the latent representation, restores missing entries with graph and dictionary priors, and trains under simulated erasures. If the claim holds, point cloud delivery over lossy networks no longer needs per-loss-rate retraining or heavy forward error correction on the main geometry bitstream.

What carries the argument

The load-bearing mechanism is the pairing of Spatial-Channel Interleaving (SCI) with Mask-Aware Graph-based Latent Restoration (MGLR). SCI reorders Morton-sorted latent features in 2x2x4 blocks so one lost packet no longer wipes out an entire channel for every anchor; instead the decoder sees scattered missing entries, each surrounded by surviving neighbors. MGLR then builds a K-nearest-neighbor graph over anchor coordinates, weights edges by mask reliability and spatial distance, and writes corrections only into masked positions, leaving received entries untouched. Two supporting modules complete the chain: Condition-Adaptive Latent Modulation (CALM) injects the loss rate into each downsampling stage through gating factors, and Dictionary-based Refinement (DBR) adds a learned cross-attention residual that pulls restored features toward canonical latent patterns.

What would settle it

Train or test with the same average loss rate but long bursts (e.g., 50–200 consecutive lost packets) and compare reconstruction PSNR against the Gilbert-Elliott results; ResPCC's margin should shrink or vanish if its restoration depends on scattered single-entry gaps rather than true burst resilience. A second check is to corrupt or drop the protected anchor and hyperprior packets and observe whether reconstruction collapses entirely.

Watch

Extended reading notes

Core claim

The central claim is that a point-cloud codec can be made intrinsically loss-resilient by conditioning encoding on the perceived packet loss rate and by restoring corrupted latent features before reconstruction. Concretely, ResPCC modulates latent generation with the loss rate, interleaves the latent tensor so channel-wise packet erasures become scattered element-wise gaps, restores those gaps with a mask-aware graph network using nearby anchors, and refines the result against a learned dictionary. Evaluated under Gilbert-Elliott bursty loss at 5% to 30%, it reports consistently higher D1/D2 PSNR than G-PCC, DPCC, PCGCv2, and OctAttention, with gains that grow as loss increases. On ShapeNet at 30% loss the reported advantage over DPCC is roughly 14.6% in D1 and 19.3% in D2, and the advantage over OctAttention exceeds 50% in both metrics.

Load-bearing premise

The scheme assumes packet loss can be captured by a single rate p that is known to both encoder and decoder and that manifests as independent erasures during training, with anchors and hyperprior protected by FEC; if real loss is burstier, p is badly estimated, or that protection fails, the reported resilience does not follow.

Editorial extensions

If this is right

  • A single ResPCC model trained once can serve a range of loss conditions, since the loss rate is an input rather than a training-scheme parameter.
  • The codec reduces the need for redundant FEC on latent data; only anchors and hyperprior are protected, and these cost a small fraction of the bitrate.
  • Performance degrades gracefully: the reported gap over DPCC widens from about 5–9% at 5% loss to about 15–19% at 30% loss on ShapeNet.
  • The approach transfers across object-like and LiDAR scenes without retraining, as shown by similar trends on ShapeNet and SemanticKITTI.

Reading between the lines

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

  • The interleaving idea likely transfers to other latent-based codecs, since the ablation shows SCI alone improves PSNR even without a restoration network.
  • A natural stress test is burst-length generalization: training uses independent erasures while testing uses Gilbert-Elliott bursts, so varying burst length at a fixed loss rate would reveal whether interleaving or restoration is doing the heavy lifting.
  • The method suggests a middle path between pure source coding and full channel coding: protect only the structural metadata and let a loss-aware decoder absorb moderate latent erasures, which could be combined with unequal error protection in practice.
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 / 5 minor

Summary. The paper proposes ResPCC, an end-to-end learned point cloud codec built on a DPCC-style backbone, with four additions for packet-loss resilience: CALM conditions the encoder on a perceived loss rate; SCI permutes latent entries so packet-level/channel-wise erasures become scattered element-wise misses; MGLR restores missing latent entries using a K-NN graph with a loss mask and reliability weighting; and DBR refines restored features against a learned codebook. The encoder and decoder are trained jointly with simulated erasures at sampled loss rates, and evaluated on ShapeNet and SemanticKITTI at 5% to 30% packet loss against G-PCC, DPCC, PCGCv2, and OctAttention. The reported results show consistent R-D gains, and the appendix provides architecture, training, and additional qualitative details.

Significance. If the empirical claims hold, ResPCC addresses a practically important gap: most learned PCC methods assume lossless transport, and the proposed mechanism (one model conditioned on loss rate plus restoration) is more attractive than per-rate retraining or external concealment. The paper also provides useful analysis of why interleaving helps (turning channel-wise extinction into scattered misses) and ships a code link and detailed appendices, which support reproducibility. The significance is currently contingent, however, on correcting internal inconsistencies in the ablation and on demonstrating that the i.i.d.-trained model is actually robust under the bursty packet-loss model used at test time.

major comments (4)
  1. [Table 1] Table 1 lists the identical configuration (DBR=no, CALM=no, MGLR=yes, Linear=no, SCI=yes) twice: Category I, third row, reports Bpp 2.653 and D2-PSNR 69.152 at PLR=5%, while Category II, first row, reports Bpp 2.325 and D2-PSNR 68.850 for the same configuration at the same PLR. Since the text states every configuration is re-trained end-to-end, these two sets of numbers cannot both describe the same model. This internal inconsistency affects the ablation conclusions for MGLR and SCI, which are key to the paper's central claim.
  2. [Sec. 5.5 and Sec. 6.1] Training simulates loss by sampling an element-wise binary mask m_i 'according to p' on the interleaved H×W×C tensor, i.e., independent Bernoulli erasures. Testing instead uses a Gilbert-Elliott model at packet level. The paper never reports the G-E transition probabilities or mean burst length, nor does it specify how packet-level losses map to entries of the interleaved tensor after SCI. Without these details and a sensitivity analysis over burst lengths, the central claim of resilience under realistic bursty loss is not established; the model may have been optimized for a corruption distribution very different from the one tested.
  3. [Sec. 6.2 and Fig. 4] All reported PSNR and Bpp values in Fig. 4 and the tables appear to be single point estimates. Since packet losses are random, D1/D2 PSNR at a given PLR is a random quantity; no error bars, confidence intervals, or multiple-loss-realization statistics are reported. The claims of 'consistently delivers superior stability and R-D performance' therefore lack statistical support. At minimum, the authors should report mean±std over multiple loss patterns and over multiple training runs, or provide a significance test.
  4. [Appendix E] Appendix E states that 'the main paper included relatively successful cases' for OctAttention and that the supplementary figures highlight its high variance. This is an explicit admission of potentially selective qualitative reporting in the main text. The authors should clarify whether the quantitative curves in Fig. 4 are averages over all patches and loss realizations and, if qualitative examples are chosen, present them without selection or state the selection criterion.
minor comments (5)
  1. [Sec. 5.1 and Sec. 6.1] ResPCC's Bpp accounting for the FEC-protected anchors and hyperprior is not specified; baselines include anchor FEC overhead in Bpp, so the comparison should state explicitly whether the same overhead is included for ResPCC.
  2. [Sec. 6.1] The G-E parameters used to generate test loss sequences should be reported, and the code release should include the exact scripts that produce both the Bernoulli training masks and the Gilbert-Elliott test masks.
  3. [Fig. 6] The caption says all variants are evaluated without the DBR module; this should also be stated in the main text to avoid confusion with Table 1.
  4. [Sec. 5.3] The symbol m_i is used both for the packet-loss mask and for entries of that mask; a single notational convention (e.g., bold M for the mask tensor and scalar entries m_c) would improve readability.
  5. [Abstract and Sec. 1] The claim of being 'the first end-to-end neural point cloud codec with intrinsic resilience' should be carefully qualified in light of prior intrinsic-resilience work in other modalities and existing point-cloud error-concealment methods; if no intra-frame point-cloud baseline exists, this should be stated explicitly rather than implied.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: ResPCC is an empirical trained-codec paper whose claims are benchmarked against external baselines, and no prediction reduces to a fitted parameter or self-citation.

full rationale

ResPCC makes no first-principles predictive claim. The central objective (Eq. 4) is a training loss that includes the learned restoration operator, and the reported numbers are measured on held-out test data against external baselines (G-PCC, DPCC, PCGCv2, OctAttention). The loss-rate conditioning uses p as an input, not as a quantity derived from the output; Fig. 6 explicitly tests sensitivity to mismatched p, so the method does not define its success metric in terms of its conditioning variable. The SCI permutation, MGLR graph aggregation, and DBR cross-attention are trainable or fixed architectural components, not renamed empirical regularities. The few self-citations (e.g., ref. [46] for the cliff effect and ref. [43] for a survey) support general context and are not load-bearing for the claimed contribution. Remaining concerns, such as missing Gilbert-Elliott burst parameters and absent error bars, are experimental-reporting gaps and correctness risks, not circularity. Therefore the circularity score is 0.

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

The paper's contribution is an architecture; it relies on standard learned-compression machinery (hyperprior, chamfer loss) and introduces hand-selected hyperparameters for loss conditioning, interleaving, and restoration. The most consequential assumption is that a single loss-rate prior describes the channel and that training erasure patterns transfer to bursty test loss.

free parameters (8)
  • Rate-distortion Lagrange multiplier λ = not reported
    Balances distortion and rate in Eq. (20). Multiple values are typically used to trace R-D curves; none are listed in the paper.
  • Density loss weight α = 1e-4
    Weight for density-aware loss in Eq. (21), chosen by hand.
  • Cardinality loss weight β = 5e-7
    Weight for cardinality loss in Eq. (21), chosen by hand.
  • Density distribution coefficient γ = 50
    Weight for mean and standard deviation terms inside density loss, Eq. (25), chosen by hand.
  • Packet loss rate sampling parameters = P(0)=0.5; Beta(α=1.1, β=2) scaled to [0, 50%]
    Training distribution for simulated packet loss, Sec. 5.5 and Sec. C.1; chosen by hand to balance packet-free and lossy conditions.
  • SCI block shape = 2×2×4
    Local block size for interleaving, Sec. 5.3; chosen by hand and not ablated.
  • MGLR neighborhood size K = 8
    Number of nearest neighbors for graph construction, Sec. C.3; also used for geometric feature extraction.
  • DBR codebook size M = 128
    Number of prototype patterns in dictionary refinement, Sec. 5.4 and Fig. 8.
assumptions (5)
  • domain assumption Packet loss over a short interval can be characterized by a single perceived loss rate p, known to both encoder and decoder.
    Used in the problem formulation (Sec. 4) and CALM conditioning; validated only partially by Fig. 6 for fixed mismatch.
  • domain assumption The transmission channel erases entries of the quantized latent multiplicatively, and mask m is known to the decoder.
    Eq. (3) and Eq. (19) model loss as a binary mask; real packet loss may corrupt bits in more complex ways.
  • domain assumption Anchors and hyperprior can be transmitted losslessly with small FEC overhead (0.38 bpp anchors, 0.4-0.7% hyperprior).
    Used in Sec. 5.1; if FEC fails, the skeleton and entropy model are corrupted.
  • standard math The latent distribution can be modeled by a variational hyperprior as in Ballé et al. [1].
    Adopted for entropy coding; standard assumption in learned compression.
  • domain assumption Training with independent random erasures generalizes to Gilbert-Elliott bursty loss at test time.
    Sec. 5.5 trains with Bernoulli-style masks; Sec. 6.1 tests with G-E bursts.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ResPCC: A Loss-Resilient Neural Point Cloud Codec over Lossy Networks." pith.science (2026). https://pith.science/paper/PV6HJ5SV

@misc{pith2026260811845,
  author       = {Pith},
  title        = {Pith review of: ResPCC: A Loss-Resilient Neural Point Cloud Codec over Lossy Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PV6HJ5SV}},
  note         = {Machine review of arXiv:2608.11845}
}
read the original abstract

Point cloud compression (PCC) is critical for efficient storage and transmission of 3D data. While recent learning-based PCC methods achieve good rate-distortion (R-D) performance, they generally rely on ideal transmission conditions. In practice, packet loss is a common issue and can severely distort latent features, causing coordinate drift and geometric degradation. To address this challenge, we present ResPCC, the first end-to-end neural point cloud codec designed to offer intrinsic resilience against data loss. Our framework is loss-rate-aware and adapts to diverse packet loss conditions. At the encoder, we introduce a Condition-Adaptive Latent Modulation (CALM) module to adjust latent feature distributions according to the perceived loss rate, as well as a Spatial-Channel Interleaving (SCI) mechanism that transforms channel-wise data extinction into spatially scattered element-wise missing patterns. At the decoder, we develop a Mask-Aware Graph-based Latent Restoration (MGLR) module, followed by a Dictionary-based Refinement (DBR) stage to recover corrupted features and align them with canonical priors. Evaluations on ShapeNet and SemanticKITTI under 5\% to 30\% packet loss rates show that ResPCC consistently delivers superior stability and R-D performance over baselines. Our framework maintains high reconstruction fidelity under lossy conditions, providing a reliable solution for 3D data transmission over practical networks. Code is available at https://github.com/starrynight314/ResPCC.

Figures

Figures reproduced from arXiv: 2608.11845 by the authors.

Figure 1
Figure 1. Impact of packet loss on the reconstruction quality [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The overall framework of ResPCC. The encoder generates anchors [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Rearranging y2𝐷 via SCI. White elements denote lost data. While the absence of SCI results in channel-wise data extinction, SCI transforms such damage into element￾wise missing values distributed across multiple channels. geometric anchors A ∈ R 𝑁𝑆 ×3 and their latent features y ∈ R 𝑁𝑆 ×𝐶 from the final downsampling stage, we first compute a permutation 𝜋 by sorting the 3D points in A according to their Morton order… view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: Comparison of R-D performance between ResPCC and baselines under packet loss rates (PLR) from 5% to 30%. (a)-(b) [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Qualitative comparison under 20% packet loss. The color bar indicates reconstruction error, where the maximum [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Sensitivity of ResPCC to prior 𝑝 on ShapeNet: matched 𝑝 vs. fixed prior (𝑝 = 5%) and w/o CALM (all variants are evaluated without DBR module). Impact of Restoration Backbone and SCI. As presented in [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Detailed architecture of the Encoder and Downsample Layer. The encoder hierarchically extracts features while the [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]
Figure 8
Figure 8. Figure 8: Detailed architecture of the Hyperprior network and the restoration pipeline. MGLR restores corrupted latents using [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]
Figure 9
Figure 9. Figure 9: Visual comparison at 5% PLR. ResPCC and baselines show relatively stable results, though our method achieves higher [PITH_FULL_IMAGE:figures/full_fig_p015_9.png]
Figure 10
Figure 10. Figure 10: Visual comparison at 10% PLR. Noticeable sparsity starts appearing in PCGCv2 and OctAttention, while ResPCC [PITH_FULL_IMAGE:figures/full_fig_p016_10.png]
Figure 11
Figure 11. Figure 11: Visual comparison at 20% PLR. OctAttention exhibits extreme instability, whereas ResPCC effectively restores the [PITH_FULL_IMAGE:figures/full_fig_p017_11.png]
Figure 12
Figure 12. Figure 12: Visual comparison at 30% PLR. Under severe loss, ResPCC is the only method that consistently preserves a recognizable [PITH_FULL_IMAGE:figures/full_fig_p018_12.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 45 canonical work pages

  1. [1]

    Johannes Ballé, David Minnen, Saurabh Singh, Sung Jin Hwang, and Nick John- ston. 2018. Variational image compression with a scale hyperprior.arXiv preprint arXiv:1802.01436(2018)

    arXiv 2018

  2. [2]

    Jens Behley, Martin Garbade, Andres Milioto, Jan Quenzel, Sven Behnke, Cyrill Stachniss, and Jurgen Gall. 2019. Semantickitti: A dataset for semantic scene understanding of lidar sequences. InProceedings of the IEEE/CVF international conference on computer vision. 9297–9307

    2019

  3. [3]

    Sourav Biswas, Jerry Liu, Kelvin Wong, Shenlong Wang, and Raquel Urtasun

    work page

  4. [4]

    Jeff Castura and Yongyi Mao. 2006. Rateless coding over fading channels.IEEE communications letters10, 1 (2006), 46–48

    work page 2006

  5. [5]

    Angel X Chang, Thomas Funkhouser, Leonidas Guibas, Pat Hanrahan, Qixing Huang, Zimo Li, Silvio Savarese, Manolis Savva, Shuran Song, Hao Su, et al

  6. [6]

    Zhili Chen, Zian Qian, Sukai Wang, and Qifeng Chen. 2022. Point cloud compres- sion with sibling context and surface priors. InEuropean Conference on Computer Vision. Springer, 744–759

    work page 2022

  7. [7]

    Yihua Cheng, Ziyi Zhang, Hanchen Li, Anton Arapin, Yue Zhang, Qizheng Zhang, Yuhan Liu, Kuntai Du, Xu Zhang, Francis Y Yan, et al. 2024. GRACE: Loss-Resilient Real-Time video through neural codecs. In21st USENIX Symposium on Networked Systems Design and Implementation (NSDI 24). 509–531

    work page 2024

  8. [8]

    Christopher Choy, JunYoung Gwak, and Silvio Savarese. 2019. 4d spatio-temporal convnets: Minkowski convolutional neural networks. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 3075–3084

    work page 2019

Show all 55 references
  1. [9]

    Mingyue Cui, Junhua Long, Mingjian Feng, Boyang Li, and Huang Kai. 2023. OctFormer: Efficient octree-based transformer for point cloud compression with local enhancement. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 37. 470–478

    2023

  2. [10]

    Olivier Devillers and P-M Gandoin. 2000. Geometric compression for interactive transmission. InProceedings Visualization 2000. VIS 2000 (Cat. No. 00CH37145). IEEE, 319–326

    2000

  3. [11]

    Chunyang Fu, Ge Li, Rui Song, Wei Gao, and Shan Liu. 2022. Octattention: Octree- based large-scale contexts model for point cloud compression. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 36. 625–633

    2022

  4. [12]

    Frank Galligan, Michael Hemmer, Ondrej Stava, Fan Zhang, and Jamieson Brettle

  5. [13]

    Linyao Gao, Tingyu Fan, Jianqiang Wan, Yiling Xu, Jun Sun, and Zhan Ma. 2021. Point cloud geometry compression via neural graph sampling. In2021 IEEE International Conference on Image Processing (ICIP). IEEE, 3373–3377

    2021

  6. [14]

    Danillo Graziosi, Ohji Nakagami, Satoru Kuma, Alexandre Zaghetto, Teruhiko Suzuki, and Ali Tabatabai. 2020. An overview of ongoing point cloud compression standardization activities: Video-based (V-PCC) and geometry-based (G-PCC). APSIPA Transactions on Signal and Information ...

    2020

  7. [15]

    André FR Guarda, Nuno MM Rodrigues, and Fernando Pereira. 2020. Adaptive deep learning-based point cloud geometry coding.IEEE Journal of Selected Topics in Signal Processing15, 2 (2020), 415–430

    2020

  8. [16]

    Yulan Guo, Hanyun Wang, Qingyong Hu, Hao Liu, Li Liu, and Mohammed Bennamoun. 2020. Deep learning for 3d point clouds: A survey.IEEE Transactions on Pattern Analysis and Machine Intelligence43, 12 (2020), 4338–4364

    2020

  9. [17]

    Gerhard Haßlinger and Oliver Hohlfeld. 2008. The Gilbert-Elliott model for packet loss in real time services on the Internet. In14th GI/ITG Conference-Measurement, Modelling and Evalutation of Computer and Communication Systems. VDE, 1–15

    2008

  10. [18]

    Yun He, Xinlin Ren, Danhang Tang, Yinda Zhang, Xiangyang Xue, and Yanwei Fu. 2022. Density-preserving deep point cloud compression. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2333–2342

    2022

  11. [19]

    I-Chun Huang, Yuang Shi, Yuan-Chun Sun, Wei Tsang Ooi, Chun-Ying Huang, and Cheng-Hsin Hsu. 2025. Composing Error Concealment Pipelines for Dy- namic 3D Point Cloud Streaming.ACM Transactions on Multimedia Computing, Communications and Applications21, 6 (2025), 1–28

    2025

  12. [20]

    Lila Huang, Shenlong Wang, Kelvin Wong, Jerry Liu, and Raquel Urtasun. 2020. Octsqueeze: Octree-structured entropy model for lidar compression. InProceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 1313–1323

    2020

  13. [21]

    Tianxin Huang and Yong Liu. 2019. 3d point cloud geometry compression on deep learning. InProceedings of the 27th ACM international conference on multimedia. 890–898

    2019

  14. [22]

    Tianxin Huang, Jiangning Zhang, Jun Chen, Zhonggan Ding, Ying Tai, Zhenyu Zhang, Chengjie Wang, and Yong Liu. 2022. 3QNet: 3D point cloud geometry quantization compression network.ACM Transactions on Graphics (TOG)41, 6 (2022), 1–13

    2022

  15. [23]

    Tzu-Kuan Hung, I-Chun Huang, Samuel Rhys Cox, Wei Tsang Ooi, and Cheng- Hsin Hsu. 2022. Error concealment of dynamic 3d point cloud streaming. In Proceedings of the 30th ACM International Conference on Multimedia. 3134–3142

    2022

  16. [24]

    2021.MPEG 3D Graphics Coding: Common test condi- tions for G-PCC

    ISO/IEC JTC1/SC29/WG7. 2021.MPEG 3D Graphics Coding: Common test condi- tions for G-PCC. Technical Report N00106. ISO/IEC

    2021

  17. [25]

    Birendra Kathariya, Li Li, Zhu Li, Jose Alvarez, and Jianle Chen. 2018. Scalable point cloud geometry coding with binary tree embedded quadtree. In2018 IEEE International Conference on Multimedia and Expo (ICME). IEEE, 1–6

    2018

  18. [26]

    Diederik P Kingma and Jimmy Ba. 2014. Adam: A method for stochastic opti- mization.arXiv preprint arXiv:1412.6980(2014)

    2014 arXiv

  19. [27]

    David JC MacKay. 2005. Fountain codes.IEE Proceedings-Communications152, 6 (2005), 1062–1068

    2005

  20. [28]

    Donald Meagher. 1982. Geometric modeling using octree encoding.Computer graphics and image processing19, 2 (1982), 129–147

    1982

  21. [29]

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, et al. 2019. Pytorch: An imperative style, high-performance deep learning library.Advances in neural information processing systems...

    2019

  22. [30]

    Maurice Quach, Giuseppe Valenzise, and Frederic Dufaux. 2019. Learning convo- lutional transforms for lossy point cloud geometry compression. In2019 IEEE international conference on image processing (ICIP). IEEE, 4320–4324

    2019

  23. [31]

    Zizheng Que, Guo Lu, and Dong Xu. 2021. Voxelcontext-net: An octree based framework for point cloud compression. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 6042–6051

    2021

  24. [32]

    Michael Rudow, Francis Y Yan, Abhishek Kumar, Ganesh Ananthanarayanan, Martin Ellis, and KV Rashmi. 2023. Tambur: Efficient loss recovery for videocon- ferencing via streaming codes. In20th USENIX Symposium on Networked Systems Design and Implementation (NSDI 23). 953–971

    2023

  25. [33]

    Ruwen Schnabel and Reinhard Klein. 2006. Octree-based Point-Cloud Compres- sion.PBG@ SIGGRAPH3, 3 (2006)

    2006

  26. [34]

    Sebastian Schwarz, Marius Preda, Vittorio Baroncini, Madhukar Budagavi, Pablo Cesar, Philip A Chou, Robert A Cohen, Maja Krivokuća, Sébastien Lasserre, Zhu Li, et al . 2018. Emerging MPEG standards for point cloud compression. IEEE Journal on Emerging and Selected Topics in Ci...

    2018

  27. [35]

    Hongwei Sha, Muchen Dong, Quanyou Luo, Ming Lu, Hao Chen, and Zhan Ma

  28. [36]

    Rui Song, Chunyang Fu, Shan Liu, and Ge Li. 2023. Efficient hierarchical en- tropy model for learned point cloud compression. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 14368–14377

    2023

  29. [37]

    Jianqiang Wang, Dandan Ding, Zhu Li, Xiaoxing Feng, Chuntong Cao, and Zhan Ma. 2022. Sparse tensor-based multiscale representation for point cloud geometry compression.IEEE Transactions on Pattern Analysis and Machine Intelligence45, 7 (2022), 9055–9071

    2022

  30. [38]

    Jianqiang Wang, Dandan Ding, Zhu Li, and Zhan Ma. 2021. Multiscale point cloud geometry compression. In2021 Data Compression Conference (DCC). IEEE, 73–82

    2021

  31. [39]

    Jianqiang Wang, Hao Zhu, Haojie Liu, and Zhan Ma. 2021. Lossy point cloud geometry compression via end-to-end learning.IEEE Transactions on Circuits and Systems for Video Technology31, 12 (2021), 4909–4923

    2021

  32. [40]

    Yifan Wang, Mufan Liu, Qi Yang, He Huang, Le Yang, and Yiling Xu. 2025. On the Efficient Adaptive Streaming of 3D Gaussian Splatting over Dynamic Networks. MM ’26, November 10–14, 2026, Rio de Janeiro, Brazil Xueqin Niu et al. IEEE Transactions on Circuits and Systems for Vide...

    2025

  33. [41]

    Louis Wiesmann, Andres Milioto, Xieyuanli Chen, Cyrill Stachniss, and Jens Behley. 2021. Deep compression for dense point cloud maps.IEEE Robotics and Automation Letters6, 2 (2021), 2060–2067

    2021

  34. [42]

    Cheng-Hao Wu, Xiner Li, Rahul Rajesh, Wei Tsang Ooi, and Cheng-Hsin Hsu

  35. [43]

    Yiling Xu, Yujie Zhang, Shuting Xia, Kaifa Yang, He Huang, Ziyu Shan, Wenjie Huang, Qi Yang, and Le Yang. 2025. Point cloud compression and objective quality assessment: A survey.arXiv preprint arXiv:2506.22902(2025)

    2025 arXiv

  36. [44]

    Wei Yan, Shan Liu, Thomas H Li, Zhu Li, Ge Li, et al . 2019. Deep autoencoder-based lossy geometry compression for point clouds.arXiv preprint arXiv:1905.03691(2019)

    2019 arXiv

  37. [45]

    Jianchao Yang, Mufan Liu, Puyue Hou, Yiling Xu, and Jun Sun. 2025. Neural Adaptive Contextual Video Streaming. InICASSP 2025-2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 1–5

    2025

  38. [46]

    Cixiao Zhang, Mufan Liu, Wenjie Huang, Yin Xu, Yiling Xu, and Dazhi He. 2025. Deep joint source-channel coding for wireless point cloud transmission. In2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 1–5

    2025

  39. [47]

    Junteng Zhang, Gexin Liu, Dandan Ding, and Zhan Ma. 2022. Transformer and upsampling-based point cloud compression. InProceedings of the 1st International Workshop on Advances in Point Cloud Compression, Processing and Analysis. 33–39

    2022

  40. [48]

    Yifan Zhang, Qingyong Hu, Guoquan Xu, Yanxin Ma, Jianwei Wan, and Yulan Guo. 2022. Not all points are equal: Learning highly efficient point-based detectors for 3d lidar point clouds. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 18953–18...

    2022

  41. [54]

    ) Encoder(B,3,N)(B,128,N) (B,64,N)(B,64,N/5)(B,64,N/25)(B,64,N/125) DownsampleBlock(B,3,𝑁!

    a feature generator that produces a dedicated loss-rate-aware featureF(𝑠) 𝑝 ∈R 𝐵×64×𝑁 𝑠 . The intermediate features from the parallel branches are first modulated by the gating factorsG(𝑠) via element-wise multiplica- tion. These modulated features are concatenated withF(𝑠) 𝑝 ...

    2026

  42. [55]

    C.2 Loss Function Coefficients The weighting coefficients for the distortion terms are consistent across both datasets to ensure a uniform optimization objective

    dataset, we increase the decay step to 15 epochs to accommodate its larger data volume. C.2 Loss Function Coefficients The weighting coefficients for the distortion terms are consistent across both datasets to ensure a uniform optimization objective. For the total distortion l...

    2026

  43. [2015]

    Shapenet: An information-rich 3d model repository.arXiv preprint arXiv:1512.03012(2015)

    2015 arXiv

  44. [2018]

    Google/draco: a library for compressing and decompressing 3d geometric meshes and point clouds.Draco: a library for compressing and decompressing 3D geometric meshes and point clouds(2018)

    2018

  45. [2020]

    Advances in Neural Information Processing Systems33 (2020), 22170–22181

    Muscle: Multi sweep compression of lidar using deep entropy models. Advances in Neural Information Processing Systems33 (2020), 22170–22181

    2020

  46. [2021]

    InPro- ceedings of the 31st ACM Workshop on Network and Operating Systems Support for Digital Audio and Video

    Dynamic 3D point cloud streaming: Distortion and concealment. InPro- ceedings of the 31st ACM Workshop on Network and Operating Systems Support for Digital Audio and Video. 98–105

  47. [2025]

    In Proceedings of the AAAI Conference on Artificial Intelligence, Vol

    Towards loss-resilient image coding for unstable satellite networks. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 39. 12506–12514

Pith tools

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