Pith. sign in

REVIEW 3 major objections 4 minor 1 cited by

GauS-SLAM: Dense RGB-D SLAM with Gaussian Surfels

T0 review · 3 major / 4 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 GauS-SLAM claims millimeter-level RGB-D tracking by replacing 3D Gaussian blobs with flat 2D Gaussian surfels and surface-aware depth rendering.

desk verdict The first coupled 2D Gaussian surfel SLAM system shows real promise on Replica, but the ScanNet++ SOTA claim rests on an unquantified re-tracking mechanism that needs disclosure before the headline numbers can be trusted. read the letter →

arxiv 2505.01934 v1 pith:FBYI3RRB submitted 2025-05-03 cs.CV

Yongxin Su , Lin Chen , Kaiting Zhang , Zhongliang Zhao , Chenfeng Hou , Ziping Yu This is my paper · ORCID
classification cs.CV
keywords denseRGB-DSLAM2DGaussiansurfelssurface-awaredepthrenderingsplattingcameratrackinggeometryconsistencylocalmapindoorreconstruction
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 tracking failures in Gaussian-based RGB-D SLAM come from geometry distortion in rendered depth, not from weak pose optimization. It claims that 3D Gaussian blobs produce multi-view-inconsistent depth and that alpha-blending across surfaces lets distant occluded surfaces corrupt the depth a tracker aligns against. Replacing them with 2D Gaussian surfels whose depth comes from ray-plane intersection, plus a surface-aware depth rendering step, removes that distortion and enables state-of-the-art localization. With this design, GauS-SLAM reports absolute trajectory error of 0.06 cm and 40.25 dB PSNR on Replica, and 0.42-0.47 cm on two ScanNet++ sequences, while running faster than its predecessor SplaTAM. A sympathetic reader would care because it suggests the scene representation itself, not the optimizer, is the deciding factor for coupled Gaussian SLAM.

What carries the argument

The central object is the 2D Gaussian surfel, a flat primitive lying on the scene's tangent planes, defined by center $\mu$, tangent vectors $e_u, e_v$, opacity, and color; its depth is the ray-plane intersection depth, called unbiased depth, rather than the projected-center depth of a 3D Gaussian ellipsoid. The load-bearing mechanism is surface-aware depth rendering: along each ray, once accumulated opacity first exceeds 0.5, the depth of every farther Gaussian is blended toward the median depth $d_m$ with weight $\beta_i = \exp(-(d_i-d_m)^2/(B\sigma_i^2))$, and rendered depth is then normalized by accumulated opacity so that occluded background surfaces cannot corrupt the foreground depth estimate. Tracking minimizes a loss over pixels with accumulated opacity above 0.9, treating pose optimization as optimization of the Gaussian primitives' relative pose, following the Gaussian-SLAM trick. The system wraps this in a local-map front-end that resets when the primitive count exceeds $\tau_l = 1.5 HW$, and a back-end that merges submaps, prunes low-opacity Gaussians, and runs bundle adjustment for global consistency. These pieces jointly eliminate the geometry distortion the paper identifies as the cause of tracking error.

What would settle it

Count the re-tracking events on ScanNet++ S1 and S2, or rerun with the re-tracking disabled: if the front-end frequently resets the local map, or if the ATE computed only over continuously tracked segments differs materially from the full numbers, the 0.42/0.47 cm values understate true drift.

Watch

Extended reading notes

Core claim

The central claim is that the geometry inconsistency of Gaussian depth rendering is what breaks frame-to-model tracking, and that a surface-based depth model fixes it. The paper diagnoses two sources: the center-depth model of 3D Gaussians, which estimates each primitive's depth from its projected center and is inconsistent across viewpoints, and the mutual interference of surfaces during depth blending, where a distant floor behind a chair backrest still contributes to the rendered depth and misaligns the frame. The proposed cure is a 2D Gaussian-surfel representation with intersection depth (unbiased depth) and a surface-aware depth rendering mechanism: each Gaussian's depth is pulled toward the median depth along the ray once accumulated opacity passes 0.5, with a weight that decays with distance, and the final depth is normalized by accumulated opacity. Tracking and mapping operate in a local map that is reset when primitive count grows, isolating visible surfaces from interference regions and bounding per-frame cost; the back-end merges local maps into the global map via co-visible submaps and bundle adjustment. On Replica the system attains 0.06 cm ATE-RMSE and 40.25 dB PSNR, and on ScanNet++ S1/S2 it reports 0.42 and 0.47 cm, which the paper presents as the best published numbers on those benchmarks.

Load-bearing premise

The ScanNet++ accuracy figures assume the re-tracking repair described in the supplementary (reset the local map and restart from the lost frame when rendered-depth error spikes) fires rarely enough that the reported ATE measures a continuous trajectory rather than a chain of recovered fragments.

Editorial extensions

If this is right

  • Other Gaussian-based SLAM trackers could adopt 2D Gaussian surfels with intersection depth to reduce pose drift, since the paper traces the dominant tracking error to depth representation rather than to the optimizer.
  • Per-frame tracking and mapping cost stays roughly bounded as the map grows, because the front-end only optimizes against a local map that is periodically reset, while the back-end handles global merging asynchronously.
  • Rendering quality improves by about 6 dB PSNR over SplaTAM on Replica even though 2DGS itself is usually weaker than 3DGS at view synthesis, which the paper credits to better surfel initialization from ground-truth depth and surface-aware depth rendering.
  • Reconstruction meshes become smoother, especially on planar regions, because 2D Gaussian surfels model surfaces directly instead of isotropic blobs.
  • The method remains sensitive to motion blur and exposure variation, and on TUM-RGBD and ScanNet its tracking gains are smaller and sometimes below methods with loop closure; the paper names this as its main limitation.

Reading between the lines

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

  • The ScanNet++ numbers likely mix continuous tracking with the re-tracking repair described in the supplementary material; the paper does not report how often the front-end flags a lost frame and resets the local map, so the 0.42/0.47 cm ATE may understate drift on the hardest segments.
  • The geometry-consistency metric used in the ablation (fix poses from the first four frames, then measure average rendered-depth L1 over 60 viewpoints) is a cheap diagnostic that any Gaussian SLAM paper could report on real sequences to separate representation error from optimizer error.
  • The depth-adjustment formula keys off the median-depth crossing and the variance of depths along each ray; synthetic scenes with layered or duplicated surfaces would provide a direct stress test, since there the index $m$ at which accumulated opacity crosses 0.5 is unstable.
  • Because surfel initialization relies on ground-truth depth and normals, porting this approach to monocular SLAM would need learned depth priors; the local-map design, however, should transfer independently of the depth-rendering change.
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

3 major / 4 minor

Summary. The paper proposes GauS-SLAM, a dense RGB-D SLAM system that represents scenes with 2D Gaussian surfels. The system uses a front-end that tracks a frame-to-model within a local map, a surface-aware depth rendering scheme to reduce multi-view geometry inconsistency, and a back-end that merges local maps and runs bundle adjustment over keyframe poses. The authors report state-of-the-art tracking on Replica and ScanNet++, with ATE-RMSE of 0.06 cm on Replica and 0.42/0.47 cm on two ScanNet++ sequences, and also report strong rendering quality. The main claimed contributions are the 2D Gaussian surfel representation for SLAM, the surface-aware depth rendering mechanism, and a local-map design that isolates visible surfaces during tracking.

Significance. If the reported results are confirmed, GauS-SLAM would be a meaningful advance in Gaussian-based dense RGB-D SLAM: it directly addresses geometry distortion in Gaussian depth rendering and demonstrates that 2D Gaussian surfels can improve both tracking and reconstruction. The ablation studies in Tables 4 and 5 support the utility of the proposed depth-rendering and local-map components, and the Replica results are consistently strong across sequences. However, the headline ScanNet++ claim relies on a re-tracking mechanism whose behavior is not quantified, and the paper contains an internal inconsistency in its rendering-performance claim. The method has no released code at submission time, and the project page is only announced, so independent verification is currently limited.

major comments (3)
  1. [Supplementary §7 (Re-tracking) and Tables 2, 10] The ScanNet++ state-of-the-art claim (Table 2: 0.42 cm on S1 and 0.47 cm on S2; Table 10: 0.31 cm average) depends on a re-tracking mechanism described only in the supplementary. This mechanism resets the local map on a lost frame, re-initializes tracking from that frame, and then extends tracking over a broader set of submaps, which is effectively a relocalization step. The supplementary states that the loss phenomenon is 'particularly observed in the S1 sequence'—the sequence for which the headline number is reported—yet the paper reports no lost-frame counts, no segment lengths, no per-segment ATE, and no ablation with re-tracking disabled. As it stands, the reported ATE may reflect a trajectory stitched from short locally accurate segments and global relocalizations rather than continuous frame-to-model odometry, so the claimed 84% improvement over LoopSplat is not yet supported.
  2. [Table 2 versus Supplementary Table 10] The evaluation protocol for ScanNet++ is not consistent across the manuscript. Table 2 states that only methods marked with '*' were evaluated on the first 250 frames, implying that SplaTAM and GauS-SLAM ran the full sequences, whereas Supplementary Table 10 says that 'we evaluate the first 250 frames for sequences' and reports all methods on that basis. Please clarify exactly which frames were used for each row in Table 2 and report full-trajectory results if they exist; without this clarification, the comparison between GauS-SLAM and the starred baselines may not be apples-to-apples.
  3. [Figure 1 and Table 1] Figure 1 claims 'GauS-SLAM's SOTA performance on the Replica dataset, achieving an ATE-RMSE of 0.06cm and 40.25 dB in rendering quality,' and the abstract emphasizes superior rendering fidelity. However, Table 1 lists Gaussian-SLAM with PSNR 42.08 dB, which is higher than the reported 40.25 dB of GauS-SLAM. Thus the rendering-quality claim is not state of the art on Replica; please correct the claim or qualify it as 'among the best' rather than SOTA.
minor comments (4)
  1. [Table 3 versus Table 11] Table 3 reports an ATE of 0.05 cm on Room0, while Table 11 reports 0.06 cm for the same sequence; please clarify whether these are different runs or a typo.
  2. [Eq. (9)] Equation (9) divides by σ_i, which can be zero when all previous depths along the ray equal d_m; please specify a numerical safeguard or a convention for this case.
  3. [Section 3.2] The median depth d_m is defined using the first Gaussian at which accumulated opacity exceeds 0.5; please state the behavior for rays whose accumulated opacity never reaches 0.5.
  4. [Section 3.5] The keyframe criterion 'proportion of newly observed scene' is not defined precisely; please provide a concrete formula or algorithm for computing this proportion.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the claimed predictions are evaluated against external benchmarks with reported ablations, and no step reduces by construction to a fitted input or self-citation.

full rationale

GauS-SLAM's central claims are supported by direct evaluation on external benchmarks (Replica, TUM-RGBD, ScanNet, ScanNet++) against published baselines, with ablations in Tables 4 and 5 isolating the contributions of depth rendering and SLAM components. Hyperparameters such as B, lambda1, lambda2, tau_k, and tau_l are fixed and reported in Section 4.1, not optimized to minimize the target ATE or PSNR metrics. The core tracking loss (Eq. 12) and mapping loss (Eq. 16) optimize poses and scene geometry against observed RGB-D data, and the ATE is then measured against ground-truth trajectories, so no reported result is defined in terms of itself. The only author self-citation, reference [3] in the related-work discussion of decoupled SLAM systems, is not load-bearing for any central claim. The supplementary re-tracking mechanism described in Section 7 is a robustness/validity concern about how lost frames are handled and how the ScanNet++ ATE should be interpreted; it is not a step in which a prediction reduces by the paper's own equations to a fitted value or to a self-citation. The stated limitation on sensitivity to motion blur and exposure variation is an acknowledged limitation, not a circular step. Overall, the derivation chain is self-contained with respect to the benchmarks and ablations presented.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The central claim rests on several hand-chosen thresholds and one ad-hoc depth-blending heuristic. All other machinery (2DGS, Gaussian splatting, NetVLAD, TSDF) is taken from prior work. No new physical entities are introduced.

free parameters (7)
  • B (depth adjustment sensitivity) = 4
    Controls down-weighting of Gaussian depths far from the median depth in Eq. (9); set by hand, not fitted.
  • lambda1, lambda2 (mapping loss weights) = 0.5, 0.1
    Weights for color/depth and depth-regularization terms in Eq. (16); chosen by hand.
  • tau_k (keyframe threshold) = 1%
    Fraction of newly observed pixels needed to trigger a keyframe; chosen by hand.
  • tau_l (local map size limit) = 1.5 x H x W
    Maximum number of Gaussian primitives per local map before it is sent to the backend; chosen by hand.
  • Per-dataset tracking/mapping hyperparameters = lr, lt, itert, iterm per Table 6
    Learning rates and iteration counts are tuned per dataset (e.g., lr=0.01, lt=0.04, itert=120 for ScanNet++).
  • Opacity thresholds for tracking mask, initialization, pruning = 0.9 / 0.6 / 0.4-0.6 / 0.05
    Used in Eq. (12), Sec. 3.4 and backend pruning; hand-set.
  • Adam optimizer betas for pose = 0.7, 0.99
    Chosen to accelerate pose convergence, Sec. 4.1.
assumptions (5)
  • domain assumption Frame-to-model tracking can be solved by optimizing the relative pose of Gaussian primitives with a fixed camera, using photometric and depth residuals (Eq. 11, Eq. 12).
    This is the standard alignment assumption in dense SLAM; it requires a reasonable initial pose and mostly static scenes.
  • domain assumption 2D Gaussian surfels give multi-view consistent depth because ray intersections with the surfel plane are unbiased surface intersections (from 2DGS [8]).
    The system inherits this property from the 2DGS representation; the paper's ablation supports it on Replica Room0 but does not prove it in general.
  • domain assumption Ground-truth depth from the RGB-D sensor is sufficiently accurate to place and orient Gaussian surfels during mapping (Eq. 13-15).
    The method unprojects pixels to initialize surfels and uses GT normals; the paper admits that poor depth quality on TUM-RGBD degrades tracking.
  • ad hoc to paper The first surface at which accumulated opacity passes 0.5 is the true visible surface, and all further Gaussians along the ray should be down-weighted (Eq. 8-9).
    This median-depth heuristic is proposed in Sec. 3.2; it can fail for transparent, reflective, or thin structures, and no failure cases are analyzed.
  • domain assumption Tracking against a bounded local map prevents outliers in the inactive global map from corrupting pose estimation, and the reset procedure preserves global consistency via backend BA.
    This is a design assumption validated by ablation F; it depends on NetVLAD co-visibility and BA solving any residual global inconsistency.

how reviews work

0 comments
Cite this review

Pith. "Pith review of GauS-SLAM: Dense RGB-D SLAM with Gaussian Surfels." pith.science (2026). https://pith.science/paper/FBYI3RRB

@misc{pith2026250501934,
  author       = {Pith},
  title        = {Pith review of: GauS-SLAM: Dense RGB-D SLAM with Gaussian Surfels},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FBYI3RRB}},
  note         = {Machine review of arXiv:2505.01934}
}
read the original abstract

We propose GauS-SLAM, a dense RGB-D SLAM system that leverages 2D Gaussian surfels to achieve robust tracking and high-fidelity mapping. Our investigations reveal that Gaussian-based scene representations exhibit geometry distortion under novel viewpoints, which significantly degrades the accuracy of Gaussian-based tracking methods. These geometry inconsistencies arise primarily from the depth modeling of Gaussian primitives and the mutual interference between surfaces during the depth blending. To address these, we propose a 2D Gaussian-based incremental reconstruction strategy coupled with a Surface-aware Depth Rendering mechanism, which significantly enhances geometry accuracy and multi-view consistency. Additionally, the proposed local map design dynamically isolates visible surfaces during tracking, mitigating misalignment caused by occluded regions in global maps while maintaining computational efficiency with increasing Gaussian density. Extensive experiments across multiple datasets demonstrate that GauS-SLAM outperforms comparable methods, delivering superior tracking precision and rendering fidelity. The project page will be made available at https://gaus-slam.github.io.

Figures

Figures reproduced from arXiv: 2505.01934 by the authors.

Figure 1
Figure 1. GauS-SLAM is a dense SLAM system using 2D Gaussian surfels[ [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Two challenges in Gaussian-based tracking methods. [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Overview of GauS-SLAM. This framework consists of a front-end that performs tracking and mapping using a single local map, and a back-end responsible for merging the local map into the global map and submap-based global optimization. 3DGS based dense SLAM 3D Gaussian primitives offer a concise and flexible ellipsoid representation compared to Point-NeRF[29]. For camera pose estimation, some 3DGS￾based approaches [3,… view at source ↗
Figures from the paper (8 more)
Figure 5
Figure 5. Figure 5: Comparison of mesh results on Replica[22]. Com￾pared to isotropic 3D Gaussians, Gaussian surfels produce smoother mesh reconstructions. Rendering and Reconstruction Performance. In the Tab. 1, we present the rendering and reconstruction perfor￾mance of GauS-SLAM on the…
Figure 4
Figure 4. Figure 4: The comparison of Rendering performance on Replica[22]. We present rendered color maps and depth error maps from 2 viewpoints to comparatively evaluate the rendering quality and geometry accuracy of different approaches. Frame 40 Frame 55 3DGS[13] 2DGS[8] GauS-SLAM(Our…
Figure 6
Figure 6. Figure 6: In the geometry consistency experiment, the error maps [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: NVS on ScanNet++[33]. Methods Metrics Novel View Training View S1 S2 Avg. S1 S2 Avg. Point-SLAM[20] PSNR [dB] ↑ 12.10 11.73 11.91 14.62 14.30 14.46 SSIM ↑ 0.31 0.26 0.28 0.35 0.41 0.38 LPIPS ↓ 0.62 0.74 0.68 0.68 0.62 0.65 SplaTAM[12] PSNR [dB] ↑ 23.99 24.84 24.41 27.8…
Figure 8
Figure 8. Figure 8: The comparison of system efficiency over time. No￾tably, GauS-SLAM performs mapping operations exclusively on keyframes, and the mapping time of non-keyframe corresponds to the recent keyframe. Start End [PITH_FULL_IMAGE:figures/full_fig_p013_8.png]
Figure 9
Figure 9. Figure 9: The comparison of trajectories with and without local map.. The local tracking method effectively avoids the influence of outlier regions in the blue area. scene without local mapping by conducting experiments on the fr3/office sequence in TUM-RGBD[24]. In this sequenc…
Figure 10
Figure 10. Figure 10: Rendering performance on TUM-RGBD[24]. Method Metric R0 R1 R2 O0 O1 O2 O3 O4 Avg. ESLAM[11] PSNR↑ 25.25 25.31 28.09 30.33 27.04 27.99 29.27 29.15 27.80 SSIM↑ 0.874 0.245 0.935 0.934 0.910 0.942 0.953 0.948 0.921 LPIPS↓ 0.315 0.296 0.245 0.213 0.254 0.238 0.186 0.210 0…
Figure 11
Figure 11. Figure 11: The comparison of mesh results in ScanNet++[33]. 5 [PITH_FULL_IMAGE:figures/full_fig_p015_11.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. $S^3$LAM: Surfel Splatting SLAM for Geometrically Accurate Tracking and Mapping

    cs.CV 2025-07 conditional novelty 5.0 of 10

    A real-time RGB-D SLAM system built on 2D Gaussian surfels with a depth-distortion adaptive rendering rule and analytic surfel pose Jacobians reports accurate geometry and robust tracking on Replica, TUM-RGBD, and ScanNet++.

Reference graph

Works this paper leans on

46 extracted references · 38 canonical work pages · cited by 1 Pith paper

  1. [1]

    Netvlad: Cnn architecture for weakly supervised place recognition

    Relja Arandjelovi ´c, Petr Gronat, Akihiko Torii, Tomas Pa- jdla, and Josef Sivic. Netvlad: Cnn architecture for weakly supervised place recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 40(6):1437–1451, 2018. 5

    work page 2018

  2. [2]

    Rp-slam: Real-time pho- torealistic slam with efficient 3d gaussian splatting, 2024

    Lizhi Bai, Chunqi Tian, Jun Yang, Siyu Zhang, Masanori Suganuma, and Takayuki Okatani. Rp-slam: Real-time pho- torealistic slam with efficient 3d gaussian splatting, 2024. 2, 3

    work page 2024

  3. [3]

    G2-mapping: General gaussian mapping for monocular, rgb-d, and lidar- inertial-visual systems

    Lin Chen, Boni Hu, Jvboxi Wang, Shuhui Bu, Guang- ming Wang, Pengcheng Han, and Jian Chen. G2-mapping: General gaussian mapping for monocular, rgb-d, and lidar- inertial-visual systems. IEEE Transactions on Automation Science and Engineering, pages 1–1, 2025. 2, 3

    work page 2025

  4. [4]

    Chang, Manolis Savva, Maciej Hal- ber, Thomas Funkhouser, and Matthias Nießner

    Angela Dai, Angel X. Chang, Manolis Savva, Maciej Hal- ber, Thomas Funkhouser, and Matthias Nießner. Scannet: Richly-annotated 3d reconstructions of indoor scenes. In Proc. Computer Vision and Pattern Recognition (CVPR), IEEE, 2017. 6, 8, 1, 2, 3

    work page 2017

  5. [5]

    evo: Python package for the evalua- tion of odometry and slam

    Michael Grupp. evo: Python package for the evalua- tion of odometry and slam. https://github.com/ MichaelGrupp/evo, 2017. 6

    2017

  6. [6]

    Rgbd gs-icp slam, 2024

    Seongbo Ha, Jiung Yeon, and Hyeonwoo Yu. Rgbd gs-icp slam, 2024. 3, 6

    work page 2024

  7. [7]

    DROID-Splat: Combining end-to-end SLAM with 3D Gaussian Splatting

    Christian Homeyer, Leon Begiristain, and Christoph Schn¨orr. Droid-splat: Combining end-to-end slam with 3d gaussian splatting. ArXiv, abs/2411.17660, 2024. 2, 3

    work page Pith review arXiv 2024

  8. [8]

    2d gaussian splatting for geometrically accu- rate radiance fields

    Binbin Huang, Zehao Yu, Anpei Chen, Andreas Geiger, and Shenghua Gao. 2d gaussian splatting for geometrically accu- rate radiance fields. In SIGGRAPH 2024 Conference Papers. Association for Computing Machinery, 2024. 1, 2, 3, 4, 6, 7, 8

    work page 2024

Show all 46 references
  1. [9]

    Photo-slam: Real-time simultaneous localization and photo- realistic mapping for monocular, stereo, and rgb-d cameras

    Huajian Huang, Longwei Li, Cheng Hui, and Sai-Kit Yeung. Photo-slam: Real-time simultaneous localization and photo- realistic mapping for monocular, stereo, and rgb-d cameras. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024. 2, 3

    2024

  2. [10]

    Di-fusion: Online implicit 3d reconstruction with deep priors

    Jiahui Huang, Shi-Sheng Huang, Haoxuan Song, and Shi- Min Hu. Di-fusion: Online implicit 3d reconstruction with deep priors. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021. 2

    2021

  3. [11]

    M. M. Johari, C. Carta, and F. Fleuret. ESLAM: Effi- cient dense slam system based on hybrid representation of signed distance fields. In Proceedings of the IEEE interna- tional conference on Computer Vision and Pattern Recogni- tion (CVPR), 2023. 2, 6, 3, 4

    2023

  4. [12]

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

    Nikhil Keetha, Jay Karhade, Krishna Murthy Jatavallabhula, Gengshan Yang, Sebastian Scherer, Deva Ramanan, and Jonathon Luiten. Splatam: Splat, track and map 3d gaussians for dense rgb-d slam. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recogniti...

    2024

  5. [13]

    3d gaussian splatting for real-time radiance field rendering

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimk ¨uhler, and George Drettakis. 3d gaussian splatting for real-time radiance field rendering. ACM Transactions on Graphics, 42 (4), 2023. 1, 4, 5, 6, 7

    2023

  6. [14]

    Hidenobu Matsuki, Riku Murai, Paul H. J. Kelly, and An- drew J. Davison. Gaussian splatting slam. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024. 1, 3, 6, 7, 2, 4

    2024

  7. [15]

    Srinivasan, Matthew Tancik, Jonathan T

    Ben Mildenhall, Pratul P. Srinivasan, Matthew Tancik, Jonathan T. Barron, Ravi Ramamoorthi, and Ren Ng. Nerf: Representing scenes as neural radiance fields for view syn- thesis. In ECCV, 2020. 2

    2020

  8. [16]

    Tard ´os

    Ra ´ul Mur-Artal and Juan D. Tard ´os. ORB-SLAM2: an open-source SLAM system for monocular, stereo and RGB- D cameras. IEEE Transactions on Robotics , 33(5):1255– 1262, 2017. 2, 6

    2017

  9. [17]

    isdf: Real-time neural signed distance fields for robot per- ception

    Joseph Ortiz, Alexander Clegg, Jing Dong, Edgar Sucar, David Novotny, Michael Zollhoefer, and Mustafa Mukadam. isdf: Real-time neural signed distance fields for robot per- ception. In Robotics: Science and Systems, 2022. 2

    2022

  10. [18]

    Rtg-slam: Real-time 3d re- construction at scale using gaussian splatting

    Zhexi Peng, Tianjia Shao, Liu Yong, Jingke Zhou, Yin Yang, Jingdong Wang, and Kun Zhou. Rtg-slam: Real-time 3d re- construction at scale using gaussian splatting. ACM Trans- actions on Graphics, 2024. 3

    2024

  11. [19]

    Nerf- slam: Real-time dense monocular slam with neural radiance fields

    Antoni Rosinol, John J Leonard, and Luca Carlone. Nerf- slam: Real-time dense monocular slam with neural radiance fields. arXiv preprint arXiv:2210.13641, 2022. 2

    2022 arXiv

  12. [20]

    Os- wald

    Erik Sandstrom, Yue Li, Luc Van Gool, and Martin R. Os- wald. Point-slam: Dense neural point cloud-based slam. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2023. 2, 6, 7, 3, 4

    2023

  13. [21]

    Splat-slam: Globally optimized rgb-only slam with 3d gaussians

    Erik Sandstr ¨om, Keisuke Tateno, Michael Oechsle, Michael Niemeyer, Luc Van Gool, Martin R Oswald, and Federico Tombari. Splat-slam: Globally optimized rgb-only slam with 3d gaussians. arXiv preprint arXiv:2405.16544, 2024. 3, 7

    2024 arXiv

  14. [22]

    Julian Straub, Thomas Whelan, Lingni Ma, Yufan Chen, Erik Wijmans, Simon Green, Jakob J. Engel, Raul Mur-Artal, Carl Ren, Shobhit Verma, Anton Clarkson, Mingfei Yan, Brian Budge, Yajie Yan, Xiaqing Pan, June Yon, Yuyang Zou, Kimberly Leon, Nigel Carter, Jesus Briales, Tyler Gi...

    1906 arXiv

  15. [23]

    A benchmark for the eval- uation of rgb-d slam systems

    J ¨urgen Sturm, Nikolas Engelhard, Felix Endres, Wolfram Burgard, and Daniel Cremers. A benchmark for the eval- uation of rgb-d slam systems. In 2012 IEEE/RSJ Interna- tional Conference on Intelligent Robots and Systems , pages 573–580, 2012. 6, 8

    2012

  16. [24]

    Sturm, N

    J. Sturm, N. Engelhard, F. Endres, W. Burgard, and D. Cre- mers. A benchmark for the evaluation of rgb-d slam systems. In Proc. of the International Conference on Intelligent Robot Systems (IROS), 2012. 6, 8, 1, 2, 3, 4

    2012

  17. [25]

    Davi- son

    Edgar Sucar, Shikun Liu, Joseph Ortiz, and Andrew J. Davi- son. imap: Implicit mapping and positioning in real-time. In 2021 IEEE/CVF International Conference on Computer Vision (ICCV), pages 6209–6218, 2021. 2 9

    2021

  18. [26]

    DROID-SLAM: Deep Visual SLAM for Monocular, Stereo, and RGB-D Cameras

    Zachary Teed and Jia Deng. DROID-SLAM: Deep Visual SLAM for Monocular, Stereo, and RGB-D Cameras. Ad- vances in neural information processing systems , 2021. 2, 3

    2021

  19. [27]

    Bovik, H.R

    Zhou Wang, A.C. Bovik, H.R. Sheikh, and E.P. Simoncelli. Image quality assessment: from error visibility to structural similarity. IEEE Transactions on Image Processing , 13(4): 600–612, 2004. 6

    2004

  20. [28]

    Schonberger, Marc Pollefeys, and Martin R

    Silvan Weder, Johannes L. Schonberger, Marc Pollefeys, and Martin R. Oswald. Neuralfusion: Online depth fusion in latent space. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 3162–3172, 2021. 2

    2021

  21. [29]

    Point-nerf: Point-based neural radiance fields

    Qiangeng Xu, Zexiang Xu, Julien Philip, Sai Bi, Zhixin Shu, Kalyan Sunkavalli, and Ulrich Neumann. Point-nerf: Point-based neural radiance fields. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5438–5448, 2022. 3

    2022

  22. [30]

    Glc-slam: Gaussian splatting slam with efficient loop closure, 2024

    Ziheng Xu, Qingfeng Li, Chen Chen, Xuefeng Liu, and Jian- wei Niu. Glc-slam: Gaussian splatting slam with efficient loop closure, 2024. 2, 3

    2024

  23. [31]

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

    Chi Yan, Delin Qu, Dan Xu, Bin Zhao, Zhigang Wang, Dong Wang, and Xuelong Li. Gs-slam: Dense visual slam with 3d gaussian splatting. In CVPR, 2024. 1, 3

    2024

  24. [32]

    V ox-fusion: Dense tracking and mapping with voxel-based neural implicit representation

    Xingrui Yang, Hai Li, Hongjia Zhai, Yuhang Ming, Yuqian Liu, and Guofeng Zhang. V ox-fusion: Dense tracking and mapping with voxel-based neural implicit representation. In 2022 IEEE International Symposium on Mixed and Aug- mented Reality (ISMAR), pages 499–507, 2022. 2

    2022

  25. [33]

    Scannet++: A high-fidelity dataset of 3d indoor scenes

    Chandan Yeshwanth, Yueh-Cheng Liu, Matthias Nießner, and Angela Dai. Scannet++: A high-fidelity dataset of 3d indoor scenes. In Proceedings of the International Confer- ence on Computer Vision (ICCV), 2023. 1, 6, 8, 2, 3, 5

    2023

  26. [34]

    Gaussian opacity fields: Efficient adaptive surface reconstruction in unbounded scenes

    Zehao Yu, Torsten Sattler, and Andreas Geiger. Gaussian opacity fields: Efficient adaptive surface reconstruction in unbounded scenes. ACM Transactions on Graphics, 2024. 3

    2024

  27. [35]

    Os- wald

    Vladimir Yugay, Yue Li, Theo Gevers, and Martin R. Os- wald. Gaussian-slam: Photo-realistic dense slam with gaus- sian splatting, 2023. 1, 3, 4, 6

    2023

  28. [36]

    3dmatch: Learning local geometric descriptors from rgb-d reconstruc- tions

    Andy Zeng, Shuran Song, Matthias Nießner, Matthew Fisher, Jianxiong Xiao, and Thomas Funkhouser. 3dmatch: Learning local geometric descriptors from rgb-d reconstruc- tions. In CVPR, 2017. 6

    2017

  29. [37]

    Rade-gs: Rasterizing depth in gaussian splatting, 2024

    Baowen Zhang, Chuan Fang, Rakesh Shrestha, Yixun Liang, Xiaoxiao Long, and Ping Tan. Rade-gs: Rasterizing depth in gaussian splatting, 2024. 3

    2024

  30. [38]

    Efros, Eli Shecht- man, and Oliver Wang

    Richard Zhang, Phillip Isola, Alexei A. Efros, Eli Shecht- man, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric.CoRR, abs/1801.03924,

    arXiv

  31. [39]

    Hi-slam2: Geometry- aware gaussian slam for fast monocular scene reconstruction,

    Wei Zhang, Qing Cheng, David Skuddis, Niclas Zeller, Daniel Cremers, and Norbert Haala. Hi-slam2: Geometry- aware gaussian slam for fast monocular scene reconstruction,

  32. [40]

    Loopsplat: Loop closure by registering 3d gaussian splats, 2024

    Liyuan Zhu, Yue Li, Erik Sandstrom, Shengyu Huang, Kon- rad Schindler, and Iro Armeni. Loopsplat: Loop closure by registering 3d gaussian splats, 2024. 6, 2, 3

    2024

  33. [41]

    Oswald, and Marc Polle- feys

    Zihan Zhu, Songyou Peng, Viktor Larsson, Weiwei Xu, Hu- jun Bao, Zhaopeng Cui, Martin R. Oswald, and Marc Polle- feys. Nice-slam: Neural implicit scalable encoding for slam. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022. 2, 6

    2022

  34. [42]

    Zwicker, H

    M. Zwicker, H. Pfister, J. van Baar, and M. Gross. Ewa splatting. IEEE Transactions on Visualization and Computer Graphics, 8(3):223–238, 2002. 4 10 GauS-SLAM: Dense RGB-D SLAM with Gaussian Surfels Supplementary Material Abstract This supplementary material encompasses an eva...

    2002

  35. [43]

    This video demonstrates the experimental results of GauS-SLAM in a challenging real-world scenario S2

    Video We submit a video GauS SLAM S2.mp4. This video demonstrates the experimental results of GauS-SLAM in a challenging real-world scenario S2. It records the back- end operation of GauS-SLAM, where the deep blue lines represent the camera trajectory, the blue frustums denote...

  36. [44]

    6 presents the principal hyperpa- rameters used in our experiments on Replica[22], TUM- RGBD[24], ScanNet[4] and ScanNet++[33]

    Implementation details Hyperparameters Tab. 6 presents the principal hyperpa- rameters used in our experiments on Replica[22], TUM- RGBD[24], ScanNet[4] and ScanNet++[33]. The learn- ing rates for camera rotation lr and translation lt during tracking are dynamically adjusted u...

  37. [45]

    Novel View Synthesis We evaluate novel view synthesis (NVS) performance on sequences S1 and S2 of the ScanNet++[33] dataset, with results presented in Tab. 7. Our method demonstrates su- perior performance over SplaTAM[12] method in both ren- dering quality and depth map accur...

  38. [46]

    11 and Tab

    Additional experiments Experiments on Replica[22] Tab. 11 and Tab. 12 present comprehensive evaluations of tracking accuracy, render- ing quality, and reconstruction performance across all se- quences in the Replica dataset. Our proposed GauS- SLAM framework demonstrates state...

Pith tools

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