pith. sign in

arxiv: 2403.11247 · v3 · pith:HYNSITQOnew · submitted 2024-03-17 · 💻 cs.CV · cs.RO

Compact 3D Gaussian Splatting For Dense Visual SLAM

Tianchen Deng , Chang Nie , Shuhong Liu , Wenhua Wu , Jianfei Yang , Shenghai Yuan , Jiuming Liu , Danwei Wang
show 1 more author
Hesheng Wang
This is my paper

Pith reviewed 2026-05-24 03:30 UTC · model grok-4.3

classification 💻 cs.CV cs.RO
keywords 3D Gaussian SplattingVisual SLAMDense ReconstructionCodebook CompressionGaussian EllipsoidsBundle AdjustmentReal-time RenderingPose Estimation
0
0 comments X

The pith

A geometry codebook plus sliding-window masking cuts redundant 3D Gaussians in SLAM while keeping reconstruction quality.

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

The paper develops a compact 3D Gaussian Splatting method for dense visual SLAM. Redundant ellipsoids are removed by a sliding window masking strategy, and the remaining geometric parameters are compressed by a codebook that exploits the observed similarity among most covariance matrices. A global bundle adjustment with reprojection loss keeps pose estimation accurate. The resulting system trains and renders faster than prior Gaussian SLAM approaches yet matches their scene quality on standard benchmarks. This directly lowers memory and storage demands that have limited real-world use of dense Gaussian mapping.

Core claim

The authors show that a single geometry codebook can compress the covariance matrices of most 3D Gaussian ellipsoids without material loss of reconstruction or pose accuracy, and that a sliding-window mask can further eliminate redundant ellipsoids, producing a SLAM pipeline whose training and rendering are materially faster while scene representation quality remains state-of-the-art.

What carries the argument

The geometry codebook that encodes similar covariance matrices of 3D Gaussian ellipsoids, paired with the sliding-window masking strategy that prunes redundant ellipsoids.

If this is right

  • Training and rendering become faster than existing 3D Gaussian SLAM systems.
  • Memory and storage costs drop because fewer and smaller Gaussian parameters are stored.
  • Pose estimation remains accurate through global bundle adjustment with reprojection loss.
  • Scene representation quality stays at state-of-the-art levels on standard benchmarks.
  • Real-time rendering of dense maps becomes feasible with lower hardware requirements.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same codebook idea could be tested on non-SLAM Gaussian reconstruction pipelines to see whether the similarity assumption holds outside mapping tasks.
  • If the compression ratio scales with scene size, the method might allow long-term mapping on memory-constrained robots without periodic map resets.
  • An adaptive codebook that updates when scene geometry changes could be a natural next step if fixed-codebook accuracy degrades over time.

Load-bearing premise

The covariance matrices of most 3D Gaussian ellipsoids are similar enough that one codebook can represent them without harming reconstruction or pose accuracy.

What would settle it

Measure reconstruction PSNR, SSIM, and absolute pose error on Replica or TUM RGB-D sequences; a statistically significant drop relative to the uncompressed baseline would falsify the claim.

Figures

Figures reproduced from arXiv: 2403.11247 by Chang Nie, Danwei Wang, Hesheng Wang, Jianfei Yang, Jiuming Liu, Shenghai Yuan, Shuhong Liu, Tianchen Deng, Wenhua Wu.

Figure 1
Figure 1. Figure 1: Our framework minimizes storage and accelerates rendering while maintaining the SOTA image reconstruction performance. The proposed [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The pipeline of our GS-based SLAM system. The input of our system is RGB-D images. We start the SLAM system by initializing the 3D [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The core distinctions between GS-based SLAM systems and [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: The KL divergence distribution of the Gaussian ellipsoids with the online training of the SLAM system on different time steps (500, 1000, [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: The left figure shows the learnable mask strategy. We perform frustum selection and sliding widow reset to remove redundant Gaussian [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: The R-VQ process to represent the scale and rotation of Gaussian ellipsoids. In the first stage, we cluster the scale and rotation vectors and [PITH_FULL_IMAGE:figures/full_fig_p006_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: The rendering visualization results on the Replica dataset [ [PITH_FULL_IMAGE:figures/full_fig_p007_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: The rendering visualization results on the TUM RGB-D dataset [ [PITH_FULL_IMAGE:figures/full_fig_p008_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: We visualized the learned codebook indices for rotation vector in the replica dataset [ [PITH_FULL_IMAGE:figures/full_fig_p010_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: We visualized the learned codebook indices for scale vector in the replica dataset [ [PITH_FULL_IMAGE:figures/full_fig_p011_10.png] view at source ↗
read the original abstract

Recent work has shown that 3D Gaussian-based SLAM enables high-quality reconstruction, accurate pose estimation, and real-time rendering of scenes. However, these approaches are built on a tremendous number of redundant 3D Gaussian ellipsoids, leading to high memory and storage costs, and slow training speed. To address the limitation, we propose a compact 3D Gaussian Splatting SLAM system that reduces the number and the parameter size of Gaussian ellipsoids. A sliding window-based masking strategy is first proposed to reduce the redundant ellipsoids. Then we observe that the covariance matrix (geometry) of most 3D Gaussian ellipsoids are extremely similar, which motivates a novel geometry codebook to compress 3D Gaussian geometric attributes, i.e., the parameters. Robust and accurate pose estimation is achieved by a global bundle adjustment method with reprojection loss. Extensive experiments demonstrate that our method achieves faster training and rendering speed while maintaining the state-of-the-art (SOTA) quality of the scene representation.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper proposes Compact 3D Gaussian Splatting for Dense Visual SLAM, introducing a sliding-window masking strategy to prune redundant 3D Gaussians and a geometry codebook to compress covariance parameters based on the observation that most ellipsoids have extremely similar geometries; pose accuracy is maintained via global bundle adjustment with reprojection loss. The central claim is that the resulting system delivers faster training and rendering while preserving SOTA reconstruction and tracking quality.

Significance. If the similarity assumption and quality claims hold under quantitative scrutiny, the work would provide a practical route to lower-memory, higher-speed dense visual SLAM, directly addressing the storage and compute overhead that currently limits 3DGS-based systems on resource-constrained platforms. The combination of masking and codebook compression is a targeted engineering contribution that could be adopted by follow-on real-time reconstruction pipelines.

major comments (2)
  1. [Abstract] Abstract: The geometry-codebook motivation states that 'the covariance matrix (geometry) of most 3D Gaussian ellipsoids are extremely similar' yet supplies no quantitative support (variance of the six covariance parameters, codebook reconstruction error, or ablation on codebook size); because this unquantified similarity is the sole justification for replacing per-Gaussian covariances without material loss in reconstruction or pose accuracy, the SOTA-quality claim rests on an unverified premise.
  2. [Experiments] Experiments (results tables): The reported maintenance of SOTA quality is presented without error bars across runs, without an ablation isolating the effect of codebook size on pose and reconstruction metrics, and without explicit statement of data-exclusion rules or baseline re-implementation details; these omissions prevent verification that localized geometric degradation is not masked by the global bundle-adjustment stage.
minor comments (2)
  1. [Method] The description of how the codebook is learned and applied during optimization would benefit from an explicit equation or short algorithm box to make the compression step reproducible.
  2. [Figures] Figure captions for the qualitative results should state the exact scenes and metrics shown so readers can directly compare against the quantitative tables.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address each major comment below and commit to revisions that strengthen the quantitative justification and experimental transparency.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The geometry-codebook motivation states that 'the covariance matrix (geometry) of most 3D Gaussian ellipsoids are extremely similar' yet supplies no quantitative support (variance of the six covariance parameters, codebook reconstruction error, or ablation on codebook size); because this unquantified similarity is the sole justification for replacing per-Gaussian covariances without material loss in reconstruction or pose accuracy, the SOTA-quality claim rests on an unverified premise.

    Authors: We agree the abstract provides no quantitative metrics for the similarity observation. The full manuscript reports overall performance but does not include the requested statistics. In revision we will add: a histogram and variance statistics for the six covariance parameters across scenes, codebook reconstruction error versus codebook size, and an ablation table showing reconstruction/pose metrics for varying codebook sizes. These additions will directly support the premise. revision: yes

  2. Referee: [Experiments] Experiments (results tables): The reported maintenance of SOTA quality is presented without error bars across runs, without an ablation isolating the effect of codebook size on pose and reconstruction metrics, and without explicit statement of data-exclusion rules or baseline re-implementation details; these omissions prevent verification that localized geometric degradation is not masked by the global bundle-adjustment stage.

    Authors: We accept that the current experiments lack these elements. The revision will add error bars from multiple runs, a dedicated ablation on codebook size reporting PSNR/SSIM/ATE, explicit statements on data splits (no test data leakage) and baseline re-implementations (official code, default hyperparameters). The new ablation will report metrics before and after codebook compression (with BA held fixed) to isolate any localized degradation. revision: yes

Circularity Check

0 steps flagged

No circularity: method rests on empirical observation and independent experiments

full rationale

The paper introduces a sliding-window masking strategy and a geometry codebook motivated by the stated observation that most covariance matrices are similar. No equations, fitted parameters, or self-citations are shown that reduce the compactness claim or SOTA-quality result to the same inputs by construction. The derivation chain consists of algorithmic choices justified by external data and benchmarks rather than self-definition or renamed predictions.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only; no explicit free parameters, axioms, or invented entities are stated beyond the core modeling choice that ellipsoids share similar covariances.

pith-pipeline@v0.9.0 · 5730 in / 1071 out tokens · 20233 ms · 2026-05-24T03:30:52.456385+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Mamba-VGGT: Persistent Long-Sequence Video Geometry Grounded Transformer via External Sliding Window Mamba Memory

    cs.CV 2026-05 unverdicted novelty 7.0

    Mamba-VGGT introduces a Sliding Window Mamba memory module and Zero-Init Spatial Memory Injector to enable persistent long-range geometric reasoning in VGGT for extended video sequences.

  2. DINO-VO: Learning Where to Focus for Enhanced State Estimation

    cs.CV 2026-04 unverdicted novelty 6.0

    DINO-VO achieves state-of-the-art monocular visual odometry accuracy and generalization by training a differentiable patch selector together with multi-task features and inverse-depth bundle adjustment.

  3. VGGT-Occ: Geometry-Grounded and Density-Aware Gated Fusion for 3D Occupancy Prediction

    cs.CV 2026-05 unverdicted novelty 5.0

    VGGT-Occ embeds geometric tokens via PA-DA and uses sequential coarse-to-fine gated fusion to reach 33.00% IoU and 21.08% mIoU on SurroundOcc-nuScenes while using only ~41M parameters in the occupancy head.

  4. The Code Whisperer: LLM and Graph-Based AI for Smell and Vulnerability Resolution

    cs.SE 2026-04 unverdicted novelty 5.0

    A hybrid graph-plus-LLM framework improves detection and repair of code smells and vulnerabilities over graph-only or LLM-only baselines on multi-language datasets.

  5. A Survey on 3D Gaussian Splatting

    cs.CV 2024-01 unverdicted novelty 2.0

    A survey compiling principles, applications, benchmarks, and challenges of 3D Gaussian Splatting for explicit 3D scene representation.

Reference graph

Works this paper leans on

38 extracted references · 38 canonical work pages · cited by 5 Pith papers · 3 internal anchors

  1. [1]

    Simultaneous localization and mapping: part i,

    H. Durrant-Whyte and T. Bailey, “Simultaneous localization and mapping: part i,” IEEE robotics & automation magazine, vol. 13, no. 2, pp. 99–110, 2006

    work page 2006

  2. [2]

    Simultaneous localization and map- ping for augmented reality,

    G. Reitmayr, T. Langlotz, D. Wagner, A. Mulloni, G. Schall, D. Schmalstieg, and Q. Pan, “Simultaneous localization and map- ping for augmented reality,” in 2010 International Symposium on Ubiquitous Virtual Reality. IEEE, 2010, pp. 5–8

    work page 2010

  3. [3]

    Orb-slam: A versatile and accurate monocular slam system,

    R. Mur-Artal, J. M. M. Montiel, and J. D. Tard ´os, “Orb-slam: A versatile and accurate monocular slam system,” IEEE T ransactions on Robotics, vol. 31, no. 5, pp. 1147–1163, 2015

    work page 2015

  4. [4]

    Orb-slam2: An open-source slam system for monocular, stereo, and rgb-d cameras,

    R. Mur-Artal and J. D. Tard ´os, “Orb-slam2: An open-source slam system for monocular, stereo, and rgb-d cameras,” IEEE T ransac- tions on Robotics , vol. 33, no. 5, pp. 1255–1262, 2017

    work page 2017

  5. [5]

    Vins-mono: A robust and versatile monocular visual-inertial state estimator,

    T. Qin, P . Li, and S. Shen, “Vins-mono: A robust and versatile monocular visual-inertial state estimator,” IEEE T ransactions on Robotics, vol. 34, no. 4, pp. 1004–1020, 2018

    work page 2018

  6. [6]

    Long-term visual simul- taneous localization and mapping: Using a bayesian persistence filter-based global map prediction,

    T. Deng, H. Xie, J. Wang, and W. Chen, “Long-term visual simul- taneous localization and mapping: Using a bayesian persistence filter-based global map prediction,” IEEE Robotics & Automation Magazine, vol. 30, no. 1, pp. 36–49, 2023

    work page 2023

  7. [7]

    Robust incremental long- term visual topological localization in changing environments,

    H. Xie, T. Deng, J. Wang, and W. Chen, “Robust incremental long- term visual topological localization in changing environments,” IEEE T ransactions on Instrumentation and Measurement , vol. 72, pp. 1–14, 2022

    work page 2022

  8. [8]

    Past, present, and future of simultaneous localization and mapping: Toward the robust- perception age,

    C. Cadena, L. Carlone, H. Carrillo, Y. Latif, D. Scaramuzza, J. Neira, I. Reid, and J. J. Leonard, “Past, present, and future of simultaneous localization and mapping: Toward the robust- perception age,” IEEE T ransactions on robotics , vol. 32, no. 6, pp. 1309–1332, 2016

    work page 2016

  9. [9]

    Dtam: Dense tracking and mapping in real-time,

    R. A. Newcombe, S. J. Lovegrove, and A. J. Davison, “Dtam: Dense tracking and mapping in real-time,” in 2011 international conference on computer vision . IEEE, 2011, pp. 2320–2327

    work page 2011

  10. [10]

    Real-time large-scale dense rgb-d slam with volumetric fusion,

    T. Whelan, M. Kaess, H. Johannsson, M. Fallon, J. J. Leonard, and J. McDonald, “Real-time large-scale dense rgb-d slam with volumetric fusion,” The International Journal of Robotics Research , vol. 34, no. 4-5, pp. 598–626, 2015

    work page 2015

  11. [11]

    Elasticfusion: Real-time dense slam and light source estimation,

    T. Whelan, R. F. Salas-Moreno, B. Glocker, A. J. Davison, and S. Leutenegger, “Elasticfusion: Real-time dense slam and light source estimation,” The International Journal of Robotics Research , vol. 35, no. 14, pp. 1697–1716, 2016

    work page 2016

  12. [12]

    imap: Implicit mapping and positioning in real-time,

    E. Sucar, S. Liu, J. Ortiz, and A. J. Davison, “imap: Implicit mapping and positioning in real-time,” in ICCV, October 2021, pp. 6229–6238

    work page 2021

  13. [13]

    Nice-slam: Neural implicit scalable encoding for slam,

    Z. Zhu, S. Peng, V . Larsson, W. Xu, H. Bao, Z. Cui, M. R. Oswald, and M. Pollefeys, “Nice-slam: Neural implicit scalable encoding for slam,” in CVPR, June 2022, pp. 12 786–12 796

    work page 2022

  14. [14]

    Eslam: Efficient dense slam system based on hybrid representation of signed distance fields,

    M. M. Johari, C. Carta, and F. Fleuret, “Eslam: Efficient dense slam system based on hybrid representation of signed distance fields,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 17 408–17 419

    work page 2023

  15. [15]

    Co-slam: Joint coordinate and sparse parametric encodings for neural real-time slam,

    H. Wang, J. Wang, and L. Agapito, “Co-slam: Joint coordinate and sparse parametric encodings for neural real-time slam,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 13 293–13 302

    work page 2023

  16. [16]

    Plgslam: Progressive neural scene represenation with local to global bundle adjustment,

    T. Deng, G. Shen, T. Qin, J. Wang, W. Zhao, J. Wang, D. Wang, and W. Chen, “Plgslam: Progressive neural scene represenation with local to global bundle adjustment,”arXiv preprint arXiv:2312.09866, 2023

    work page arXiv 2023

  17. [17]

    3d gaussian splatting for real-time radiance field rendering,

    B. Kerbl, G. Kopanas, T. Leimk ¨uhler, and G. Drettakis, “3d gaussian splatting for real-time radiance field rendering,” ACM T ransactions on Graphics, vol. 42, no. 4, 2023

    work page 2023

  18. [18]

    Splatam: Splat, track & map 3d gaussians for dense rgb-d slam,

    N. Keetha, J. Karhade, K. M. Jatavallabhula, G. Yang, S. Scherer, D. Ramanan, and J. Luiten, “Splatam: Splat, track & map 3d gaussians for dense rgb-d slam,” arXiv preprint arXiv:2312.02126 , 2023

    work page arXiv 2023

  19. [19]

    Gs- slam: Dense visual slam with 3d gaussian splatting,

    C. Yan, D. Qu, D. Wang, D. Xu, Z. Wang, B. Zhao, and X. Li, “Gs- slam: Dense visual slam with 3d gaussian splatting,”arXiv preprint arXiv:2311.11700, 2023

    work page arXiv 2023

  20. [20]

    Gaussian-slam: Photo-realistic dense slam with gaussian splatting,

    V . Yugay, Y. Li, T. Gevers, and M. R. Oswald, “Gaussian-slam: Photo-realistic dense slam with gaussian splatting,” arXiv preprint arXiv:2312.10070, 2023

    work page arXiv 2023

  21. [21]

    Gaussian splatting slam,

    H. Matsuki, R. Murai, P . H. Kelly, and A. J. Davison, “Gaussian splatting slam,” arXiv preprint arXiv:2312.06741 , 2023

    work page arXiv 2023

  22. [22]

    Kinectfusion: real-time 3d reconstruction and interaction using a moving depth camera,

    S. Izadi, D. Kim, O. Hilliges, D. Molyneaux, R. Newcombe, P . Kohli, J. Shotton, S. Hodges, D. Freeman, A. Davison et al. , “Kinectfusion: real-time 3d reconstruction and interaction using a moving depth camera,” in Proceedings of the 24th annual ACM symposium on User interface software and technology , 2011, pp. 559– 568

    work page 2011

  23. [23]

    Droid-slam: Deep visual slam for monocular, stereo, and rgb-d cameras,

    Z. Teed and J. Deng, “Droid-slam: Deep visual slam for monocular, stereo, and rgb-d cameras,” Advances in neural information process- ing systems, vol. 34, pp. 16 558–16 569, 2021

    work page 2021

  24. [24]

    Raft: Recurrent all-pairs field transforms for optical flow,

    Z. Teed and J. Deng, “Raft: Recurrent all-pairs field transforms for optical flow,” in Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part II 16 . Springer, 2020, pp. 402–419

    work page 2020

  25. [25]

    Codeslam — learning a compact, optimisable representation for dense visual slam,

    M. Bloesch, J. Czarnowski, R. Clark, S. Leutenegger, and A. J. Davi- son, “Codeslam — learning a compact, optimisable representation for dense visual slam,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , June 2018

    work page 2018

  26. [26]

    Deep patch visual odometry,

    Z. Teed, L. Lipson, and J. Deng, “Deep patch visual odometry,” Advances in Neural Information Processing Systems , vol. 36, 2024

    work page 2024

  27. [27]

    Nerf: Representing scenes as neural radiance fields for view synthesis,

    B. Mildenhall, P . P . Srinivasan, M. Tancik, J. T. Barron, R. Ra- mamoorthi, and R. Ng, “Nerf: Representing scenes as neural radiance fields for view synthesis,” in ECCV, 2020

    work page 2020

  28. [28]

    Vox-fusion: Dense tracking and mapping with voxel-based neural implicit representation,

    X. Yang, H. Li, H. Zhai, Y. Ming, Y. Liu, and G. Zhang, “Vox-fusion: Dense tracking and mapping with voxel-based neural implicit representation,” in 2022 IEEE International Symposium on Mixed and Augmented Reality (ISMAR) , 2022, pp. 499–507

    work page 2022

  29. [29]

    Point- slam: Dense neural point cloud-based slam,

    E. Sandstr ¨om, Y. Li, L. Van Gool, and M. R. Oswald, “Point- slam: Dense neural point cloud-based slam,” in Proceedings of the 13 IEEE/CVF International Conference on Computer Vision , 2023, pp. 18 433–18 444

    work page 2023

  30. [30]

    Compact 3d gaussian representation for radiance field,

    J. C. Lee, D. Rho, X. Sun, J. H. Ko, and E. Park, “Compact 3d gaussian representation for radiance field,” arXiv preprint arXiv:2311.13681, 2023

    work page arXiv 2023

  31. [31]

    Compact 3d scene representation via self-organizing gaussian grids,

    W. Morgenstern, F. Barthel, A. Hilsmann, and P . Eisert, “Compact 3d scene representation via self-organizing gaussian grids,” arXiv preprint arXiv:2312.13299, 2023

    work page arXiv 2023

  32. [32]

    Lightgaus- sian: Unbounded 3d gaussian compression with 15x reduction and 200+ fps,

    Z. Fan, K. Wang, K. Wen, Z. Zhu, D. Xu, and Z. Wang, “Lightgaus- sian: Unbounded 3d gaussian compression with 15x reduction and 200+ fps,” arXiv preprint arXiv:2311.17245 , 2023

    work page arXiv 2023

  33. [33]

    Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computation

    Y. Bengio, N. L ´eonard, and A. Courville, “Estimating or propagat- ing gradients through stochastic neurons for conditional compu- tation,” arXiv preprint arXiv:1308.3432 , 2013

    work page internal anchor Pith review Pith/arXiv arXiv 2013

  34. [34]

    Soundstream: An end-to-end neural audio codec,

    N. Zeghidour, A. Luebs, A. Omran, J. Skoglund, and M. Tagliasacchi, “Soundstream: An end-to-end neural audio codec,” IEEE/ACM T ransactions on Audio, Speech, and Language Processing, vol. 30, pp. 495–507, 2021

    work page 2021

  35. [35]

    High Fidelity Neural Audio Compression

    A. D ´efossez, J. Copet, G. Synnaeve, and Y. Adi, “High fidelity neural audio compression,” arXiv preprint arXiv:2210.13438 , 2022

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  36. [36]

    The Replica Dataset: A Digital Replica of Indoor Spaces

    J. Straub, T. Whelan, L. Ma, Y. Chen, E. Wijmans, S. Green, J. J. Engel, R. Mur-Artal, C. Ren, S. Verma et al. , “The replica dataset: A digital replica of indoor spaces,” arXiv preprint arXiv:1906.05797, 2019

    work page internal anchor Pith review Pith/arXiv arXiv 1906

  37. [37]

    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,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , July 2017

    work page 2017

  38. [38]

    A benchmark for the evaluation of rgb-d slam systems,

    J. Sturm, N. Engelhard, F. Endres, W. Burgard, and D. Cremers, “A benchmark for the evaluation of rgb-d slam systems,” in 2012 IEEE/RSJ International Conference on Intelligent Robots and Systems , 2012, pp. 573–580

    work page 2012