Pith. sign in

REVIEW 4 major objections 6 minor 38 references

R3GS: Gaussian Splatting for Robust Reconstruction and Relocalization in Unconstrained Image Collections

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read R3GS is a 3D Gaussian Splatting framework that jointly handles appearance variation, transient objects, and sky in unconstrained photo collections, improving novel-view synthesis and relocalization while keeping real-time rendering.

desk verdict Useful engineering combination, but the SOTA claim doesn't survive contact with Table I. read the letter →

arxiv 2505.15294 v1 pith:ZDO33NBA submitted 2025-05-21 cs.CV cs.GRcs.LGcs.RO

Xu yan , Zhaohui Wang , Rong Wei , Jingbo Yu , Dong Li , Xiangde Liu This is my paper
classification cs.CVcs.GRcs.LGcs.RO
keywords 3DGaussianSplattingnovelviewsynthesisrelocalizationunconstrainedimagecollectionstransientobjectremovalskyhandlingappearancemodelingphotometricposeoptimization
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

This paper argues that 3D Gaussian Splatting can be made to work on unconstrained photo collections—tourist snapshots with changing daylight, passing pedestrians, cars, posters, and bright skies—without giving up real-time rendering. R3GS, the proposed framework, attaches to each Gaussian anchor a global per-image appearance feature from a CNN plus a local feature from a multiresolution hash grid, then predicts color, opacity, and covariance with shallow MLPs. A fine-tuned lightweight human detector supplies a visibility map that masks transient objects during optimization, while a large-radius sky sphere with a depth prior keeps the sky from collapsing into the foreground. The paper reports state-of-the-art novel-view-synthesis numbers on three Phototourism landmarks and a large drop in relocalization error versus plain 3DGS, at training times and frame rates close to the original method. If correct, R3GS makes high-quality reconstruction and pose estimation practical for large, messy real-world image collections.

What carries the argument

The load-bearing object is the hybrid anchor: per-anchor, a 32-dimensional local feature is queried from a multiresolution hash grid and concatenated with a 64-dimensional global appearance feature extracted from the reference image by a CNN encoder; three shallow MLPs map the concatenation to the Gaussian attributes (opacity, color, covariance). Around this core sit two supporting mechanisms. The first is a visibility map: a pretrained LRASPP-MobileNet human detector, fine-tuned for 20,000–25,000 iterations on the scene, is used to mask transient foreground objects while leaving the sky untouched. The second is a sky sphere: 20,000 Gaussians sampled by Fibonacci spherical mapping on a sphere of radius ten times the scene radius, with fixed positions, a OneFormer sky mask, and a depth prior that keeps sky geometry at infinity. Relocalization rests on CosPlace: the query's place-recognition descriptor is compared with a bank of training descriptors, the three nearest neighbors' poses are combined with distance-inverse weights to initialize the camera, and photometric loss refines the pose with Gaussian splatting gradients.

What would settle it

On a Phototourism-style scene, take held-out test images that do contain transient objects with manually labeled masks, and compare rendering and relocalization with and without R3GS's visibility map; if novel-view PSNR/LPIPS on the transient-containing frames does not improve over the no-mask baseline, or if the fine-tuned detector's IoU on non-human transient classes falls well below its human-class IoU, then the transient-generalization claim is contradicted. A cheaper check is to report the visibility map's precision and recall on held-out transients, which the paper currently omits.

Watch

Extended reading notes

Core claim

The central claim is that the three long-standing failure modes of 3DGS on in-the-wild data—appearance variation, transient objects, and sky—can be handled by one hybrid architecture without sacrificing the method's speed or memory footprint. The scene is represented by anchors, each carrying a local hash-grid code and a per-image CNN appearance code, feeding three shallow MLPs that emit Gaussian color, opacity, and covariance. Transient content is suppressed by a visibility map produced by a fine-tuned LRASPP-MobileNet detector, and the sky is rendered by Gaussians pinned to a sphere of radius ten times the scene radius, with a depth prior preventing foreground-sky depth collisions. For relocalization, an appearance-conditioned standard 3DGS model is rendered from a query image, initialized with a CosPlace-weighted pose from the three nearest training views, and refined by photometric loss. The result, the authors claim, is state-of-the-art PSNR, SSIM, and LPIPS on Brandenburg Gate, Sacre Coeur, and Trevi Fountain, and ATE of 0.101, 0.021, and 0.367 respectively against a 3DGS baseline, while rendering at 239 FPS and training in roughly 0.31 hours per 10,000 iterations.

Load-bearing premise

The load-bearing premise is that fine-tuning a human detector on the training scene yields a visibility map that correctly masks every kind of transient object—people, cars, posters, banners—and that the map is accurate enough to exclude those pixels during optimization without erasing useful scene content.

Editorial extensions

If this is right

  • On the three Phototourism scenes, R3GS exceeds NeRF-W, Ha-NeRF, CR-NeRF, WildGaussians, and the downsampled GS-W on PSNR, SSIM, and LPIPS, while rendering at 239 FPS without caching.
  • The same pipeline estimates camera poses with ATE of 0.101 (Brandenburg Gate), 0.021 (Sacre Coeur), and 0.367 (Trevi Fountain), a large improvement over the 3DGS baseline (0.401, 0.381, 1.377).
  • Because the appearance code is per-image and the exported model is a standard 3DGS point cloud, the scene can be rendered with a chosen appearance at a given pose without storing per-Gaussian high-dimensional features.
  • Masking transient objects improves visual quality in training views, and applying the mask only to the foreground prevents clouds from being misclassified as transient.
  • Fixed sky-Gaussian positions plus a depth prior eliminate the depth overlap between sky and foreground that causes floaters.

Reading between the lines

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

  • A natural extension the paper does not test is whether the fine-tuned detector transfers to a different landmark scene; the paper reports no mask-accuracy numbers, so held-out transient categories could behave worse than the training-time masks.
  • The CosPlace-based pose initialization only needs a feature bank of training poses, so the same relocalization idea could scale to very large collections where full bundle adjustment is impractical, assuming the three-nearest-neighbor retrieval is geographically reliable.
  • Because each image supplies its own appearance code, the exported 3DGS model could in principle be re-rendered with a different appearance code to simulate time-of-day or weather changes, a capability the paper notes but does not demonstrate.
  • The visibility map is applied only to the foreground so clouds are not masked; a direct test of the sky-handling claim would be to compare floaters and depth consistency on scenes with partial cloud cover against a no-sky-prior baseline.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The paper proposes R3GS, a 3D Gaussian Splatting framework for unconstrained photo collections. It combines per-image CNN appearance features with per-anchor multiresolution hash-grid features to predict Gaussian attributes, fine-tunes a lightweight human detector to produce visibility maps for masking transient objects, handles sky regions by constraining sky Gaussians to a large-radius sphere with a depth prior, and introduces a relocalization procedure that initializes poses via CosPlace retrieval and refines them with photometric loss. Experiments on three Phototourism scenes report novel-view-synthesis quality (Table I), relocalization accuracy (Table II), ablations (Table III), and training/rendering speed.

Significance. If the claims were fully supported, the hybrid anchor representation and the relocalization-by-appearance-tuning pipeline would be useful contributions to in-the-wild 3DGS, and the paper demonstrates real-time rendering with competitive NVS numbers. The paper also states that code will be open-sourced, which would aid reproducibility. However, the current quantitative evidence does not support the unconditional state-of-the-art claim: Table I shows GS-W outperforming R3GS on Brandenburg Gate and on Trevi Fountain LPIPS, the sky-handling ablation has an unexplained SSIM regression, and the visibility-map generalization claim lacks mask-level evaluation. The approach is promising, but the claims need to be aligned with the evidence before the paper is publishable.

major comments (4)
  1. [Section IV-B, Table I] The text states 'Our method outperforms all concurrent works in PSNR, SSIM, and LPIPS' and the caption of Table I says 'We significantly outperform all baseline methods', but Table I contradicts this. On Brandenburg Gate, GS-W scores 28.39 PSNR / 0.9306 SSIM / 0.0833 LPIPS versus R3GS's 27.02 / 0.9240 / 0.1317, and on Trevi Fountain GS-W has a better LPIPS (0.1640 versus 0.1950). The footnote that GS-W downsamples input by 2x and would overflow memory at full resolution may justify excluding it from a full-resolution comparison, but then the SOTA claim must be explicitly restricted to full-resolution methods. As written, the abstract, introduction, and Section IV-B make an unconditional claim that the table does not support; please rephrase the claim or add a full-resolution comparison with GS-W if feasible.
  2. [Section IV-C, Table III] Table III shows the full model's SSIM on Brandenburg Gate (0.9240) is lower than the no-sky variant (0.9421), while the text says removing the sky module causes the visibility map to treat the sky as transient and disrupts sky learning. The paper does not explain this metric regression. Since sky handling is a stated contribution, either provide an explanation (for example, if the sky mask incorrectly masks foreground pixels or the depth prior biases SSIM) or temper the claim that sky handling improves rendering fidelity; without this, the benefit of the sky module is not established.
  3. [Section III-C, Section IV-A] The transient-object removal module is a core contribution but is underspecified. The paper states that a pre-trained LRASPP-MobileNet human detector is fine-tuned for 20,000-25,000 iterations in Section III-C, yet Section IV-A says fine-tuning converges after 5,000 iterations, and the supervision signal for this fine-tuning is never stated. The claim that the visibility map 'efficiently generalizes to other transient objects (such as posters, banners, and cars)' is supported only by final rendering metrics; no mask accuracy, intersection-over-union, or qualitative mask statistics are reported. Because Section IV-C notes that the test set contains no transient objects, the NVS metrics cannot demonstrate that transient handling improves test-time rendering. Please specify the fine-tuning protocol, report mask-level metrics if possible, and discuss the generalization evidence.
  4. [Section IV-B, Table II] The relocalization experiment compares only with a 3DGS baseline, which is insufficient to support the claim of 'robust relocalization' or 'superior relocalization accuracy'. Established pose-estimation and relocalization baselines should be included, or the claim should be limited to 'improves over direct 3DGS pose optimization'. In addition, on Brandenburg Gate R3GS's rotation distance (0.582) is slightly worse than 3DGS's (0.576), and the paper does not discuss this. The ATE improvements are large, but the single-baseline comparison and the unaddressed rotation regression weaken the conclusion.
minor comments (6)
  1. [Abstract, Section III] There are numerous typos, including 'ffne-tune' in the abstract, 'golobal' in Section III-B, and 'adress' in the introduction; the paper needs a careful proofread.
  2. [Section IV-B, Relocalization] The statement that the highest and lowest ATE values are excluded in each group is not accompanied by the number of test images or the variance of the remaining values, making it hard to judge the stability of the reported improvements.
  3. [Equation (4)] Equation (4) defines alpha using sigma and delta, but these quantities are not defined in the preliminaries; in standard 3DGS the opacity alpha is a learnable attribute, so the equation should be aligned with the notation used in the rest of the paper.
  4. [Equation (10)] Equation (10) for ATE is malformed: the square root appears to apply to an underspecified sum, and no averaging or mean operation is shown; please rewrite the ATE definition clearly.
  5. [References, [38]] Reference [38] is cited for COLMAP poses, but the cited title is 'Colmap: A memory-efficient occupancy grid mapping framework', which is a different work; the correct COLMAP reference should be used.
  6. [Table II] In Table II the first row reads '0.4010.576' without a clear separator between the ATE and Rot Dist values; please format the numbers consistently.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: R3GS's rendering and relocalization claims are evaluated on held-out test images and external baselines; no load-bearing step reduces to its own inputs.

full rationale

The paper's derivation chain does not exhibit any of the enumerated circularity patterns. The hybrid representation combines a CNN global feature with hash-grid local features and MLP-predicted Gaussian attributes; this is a standard encoder-decoder construction trained with photometric losses against held-out test views, so the NVS numbers in Table I are not re-encodings of fitted parameters. The visibility map is produced by fine-tuning a pretrained LRASPP-MobileNet detector; although the fine-tuning supervision is not specified and the test set contains no transient objects, the mask is an external pretrained network output used during optimization rather than a quantity defined to match the rendering metrics. The sky handling uses OneFormer sky masks plus a fixed-radius sphere and depth-regularization losses, none of which encode the reported PSNR/SSIM/LPIPS values. The relocalization pipeline initializes poses from CosPlace retrieval over training images and then optimizes a photometric loss against the test image with the Gaussians fixed; this is genuine test-time optimization, not a fitted input renamed as a prediction. The internal contradiction in Section IV-B, where the text claims R3GS outperforms all concurrent works while Table I shows GS-W higher on Brandenburg Gate PSNR/SSIM and Trevi Fountain LPIPS, is an accuracy/correctness concern about the SOTA claim, not a circularity. No self-citation chain is load-bearing, and no uniqueness theorem or ansatz is smuggled via citation. The paper is self-contained against external benchmarks, so the appropriate finding is no significant circularity.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The system rests on established neural rendering components and a set of hand-chosen hyperparameters; no new physical entities are introduced. The most fragile assumptions are the generalization of the fine-tuned visibility map and the photometric consistency of appearance-conditioned rendering, neither of which is directly validated.

free parameters (6)
  • Sky sphere radius factor = 10x scene radius
    Fixed by hand in Section III-D; affects how far the sky sphere is placed and hence artifact behavior.
  • Scene radius ratio = 0.97 of L2 norm of centered SfM points
    Used in Section III-D to define the scene and sky sphere; arbitrary robust choice.
  • Number of sky anchors = 20,000
    Fibonacci-sphere sample count chosen in Section III-D; density of sky Gaussians is a user choice.
  • Visibility map fine-tuning schedule = 5000 to 25000 iterations (text varies)
    Section III-C says 20,000 to 25,000, Section IV-A says 5000; inconsistent, and the supervision signal is not specified.
  • Number of pose seed images = 3 nearest neighbors
    CosPlace retrieval count in Eqs. 6 to 8; impacts initialization quality.
  • Appearance and hash feature dimensions = 64-dim CNN, 32-dim hash
    Chosen in Section IV-A; not justified.
assumptions (5)
  • domain assumption After appearance conditioning, the rendered image and the query image are photometrically consistent for pose optimization.
    Section III-E relies on this to justify direct L1 photometric loss; if appearance modeling fails to account for lighting, pose optimization can diverge.
  • domain assumption The fine-tuned human detector's visibility map accurately labels transient objects in the foreground.
    Section III-C uses this mask to downweight transient pixels; no quantitative mask accuracy is given.
  • domain assumption CosPlace feature distance correlates with spatial proximity of cameras.
    Section III-E uses the three nearest neighbors in feature space to interpolate an initial pose (Eqs. 6 to 8).
  • standard math The Gaussian rasterizer can backpropagate gradients with respect to depth and alpha for the sky depth constraint.
    Section III-D invokes a modified rasterizer without code release; the reader must trust the forward/backward behavior.
  • standard math Standard 3DGS differentiable rasterization equations (Eqs. 1 to 5) hold as in reference [1].
    Used throughout as background.

how reviews work

0 comments
Cite this review

Pith. "Pith review of R3GS: Gaussian Splatting for Robust Reconstruction and Relocalization in Unconstrained Image Collections." pith.science (2026). https://pith.science/paper/ZDO33NBA

@misc{pith2026250515294,
  author       = {Pith},
  title        = {Pith review of: R3GS: Gaussian Splatting for Robust Reconstruction and Relocalization in Unconstrained Image Collections},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZDO33NBA}},
  note         = {Machine review of arXiv:2505.15294}
}
read the original abstract

We propose R3GS, a robust reconstruction and relocalization framework tailored for unconstrained datasets. Our method uses a hybrid representation during training. Each anchor combines a global feature from a convolutional neural network (CNN) with a local feature encoded by the multiresolution hash grids [2]. Subsequently, several shallow multi-layer perceptrons (MLPs) predict the attributes of each Gaussians, including color, opacity, and covariance. To mitigate the adverse effects of transient objects on the reconstruction process, we ffne-tune a lightweight human detection network. Once ffne-tuned, this network generates a visibility map that efffciently generalizes to other transient objects (such as posters, banners, and cars) with minimal need for further adaptation. Additionally, to address the challenges posed by sky regions in outdoor scenes, we propose an effective sky-handling technique that incorporates a depth prior as a constraint. This allows the inffnitely distant sky to be represented on the surface of a large-radius sky sphere, signiffcantly reducing ffoaters caused by errors in sky reconstruction. Furthermore, we introduce a novel relocalization method that remains robust to changes in lighting conditions while estimating the camera pose of a given image within the reconstructed 3DGS scene. As a result, R3GS significantly enhances rendering ffdelity, improves both training and rendering efffciency, and reduces storage requirements. Our method achieves state-of-the-art performance compared to baseline methods on in-the-wild datasets. The code will be made open-source following the acceptance of the paper.

Figures

Figures reproduced from arXiv: 2505.15294 by the authors.

Figure 1
Figure 1. Given an unconstrained image collection as input, [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Proposed R3GS model architecture-We begin with a reference view [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Visual comparison of rendering quality between different approaches. Red and green crops emphasize geometry [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Ablation study results: (a) Ground truth; (b) rendered result; (c) rendered result without the visibility map; (d) [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

38 extracted references · 24 canonical work pages

  1. [1]

    3D Gaussian Splatting for Real-Time Radiance Field Rendering,

    B. Kerbl, G. Kopanas, T. Leimkuehler, and G. Drettakis, “3D Gaussian Splatting for Real-Time Radiance Field Rendering,”ACM Transactions on Graphics, vol. 42, pp. 1–14, Aug. 2023

    work page 2023

  2. [2]

    Instant Neural Graphics Primitives with a Multiresolution Hash Encoding,

    T. M ¨uller, A. Evans, C. Schied, and A. Keller, “Instant Neural Graphics Primitives with a Multiresolution Hash Encoding,”ACM Transactions on Graphics, vol. 41, pp. 1–15, July 2022

    work page 2022

  3. [3]

    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,” Aug. 2020

    work page 2020

  4. [4]

    Nerf in the wild: Neural radiance fields for unconstrained photo collections,

    R. Martin-Brualla, N. Radwan, M. S. Sajjadi, J. T. Barron, A. Doso- vitskiy, and D. Duckworth, “Nerf in the wild: Neural radiance fields for unconstrained photo collections,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 7210–7219, 2021

    work page 2021

  5. [5]

    Hallucinated neural radiance fields in the wild,

    X. Chen, Q. Zhang, X. Li, Y . Chen, Y . Feng, X. Wang, and J. Wang, “Hallucinated neural radiance fields in the wild,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 12943–12952, 2022

    work page 2022

  6. [6]

    Cross-ray neural radiance fields for novel-view synthesis from unconstrained image collections,

    Y . Yang, S. Zhang, Z. Huang, Y . Zhang, and M. Tan, “Cross-ray neural radiance fields for novel-view synthesis from unconstrained image collections,” inProceedings of the IEEE/CVF International Conference on Computer Vision, pp. 15901–15911, 2023

    work page 2023

  7. [7]

    Fastnerf: High-fidelity neural rendering at 200fps,

    S. J. Garbin, M. Kowalski, M. Johnson, J. Shotton, and J. Valentin, “Fastnerf: High-fidelity neural rendering at 200fps,” inProceedings of the IEEE/CVF international conference on computer vision, pp. 14346–14355, 2021

    work page 2021

  8. [8]

    Instant neural graphics primitives with a multiresolution hash encoding,

    T. M ¨uller, A. Evans, C. Schied, and A. Keller, “Instant neural graphics primitives with a multiresolution hash encoding,”ACM transactions on graphics (TOG), vol. 41, no. 4, pp. 1–15, 2022

    2022

Show all 38 references
  1. [9]

    Structure-from-motion revis- ited,

    J. L. Schonberger and J.-M. Frahm, “Structure-from-motion revis- ited,” inProceedings of the IEEE conference on computer vision and pattern recognition, pp. 4104–4113, 2016

    2016

  2. [10]

    Oneformer: One transformer to rule universal image segmentation,

    J. Jain, J. Li, M. T. Chiu, A. Hassani, N. Orlov, and H. Shi, “Oneformer: One transformer to rule universal image segmentation,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 2989–2998, 2023

    2023

  3. [11]

    Scaffold-gs: Structured 3d gaussians for view-adaptive rendering,

    T. Lu, M. Yu, L. Xu, Y . Xiangli, L. Wang, D. Lin, and B. Dai, “Scaffold-gs: Structured 3d gaussians for view-adaptive rendering,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 20654–20664, 2024

    2024

  4. [12]

    Vastgaussian: Vast 3d gaussians for large scene reconstruction,

    J. Lin, Z. Li, X. Tang, J. Liu, S. Liu, J. Liu, Y . Lu, X. Wu, S. Xu, Y . Yan,et al., “Vastgaussian: Vast 3d gaussians for large scene reconstruction,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 5166–5175, 2024

    2024

  5. [13]

    Pygs: Large-scale scene representation with pyramidal 3d gaussian splatting,

    Z. Wang and D. Xu, “Pygs: Large-scale scene representation with pyramidal 3d gaussian splatting,”arXiv preprint arXiv:2405.16829, 2024

    2024 arXiv

  6. [14]

    Gs-lrm: Large reconstruction model for 3d gaussian splatting,

    K. Zhang, S. Bi, H. Tan, Y . Xiangli, N. Zhao, K. Sunkavalli, and Z. Xu, “Gs-lrm: Large reconstruction model for 3d gaussian splatting,”arXiv preprint arXiv:2404.19702, 2024

    2024 arXiv

  7. [15]

    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,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 21357– 21366, 2024

    2024

  8. [16]

    Rtg-slam: Real-time 3d reconstruction at scale using gaussian splatting,

    Z. Peng, T. Shao, L. Yong, J. Zhou, Y . Yang, J. Wang, and K. Zhou, “Rtg-slam: Real-time 3d reconstruction at scale using gaussian splatting,” 2024

    2024

  9. [17]

    Gaussian splatting slam,

    H. Matsuki, R. Murai, P. H. Kelly, and A. J. Davison, “Gaussian splatting slam,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 18039–18048, 2024

    2024

  10. [18]

    Drivinggaussian: Composite gaussian splatting for surrounding dy- namic autonomous driving scenes,

    X. Zhou, Z. Lin, X. Shan, Y . Wang, D. Sun, and M.-H. Yang, “Drivinggaussian: Composite gaussian splatting for surrounding dy- namic autonomous driving scenes,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 21634– 21643, 2024

    2024

  11. [19]

    Gaussianbev: 3d gaussian representation meets perception models for bev segmentation,

    F. Chabot, N. Granger, and G. Lapouge, “Gaussianbev: 3d gaussian representation meets perception models for bev segmentation,” 2024

    2024

  12. [20]

    Gaussianformer: Scene as gaussians for vision-based 3d semantic occupancy predic- tion,

    Y . Huang, W. Zheng, Y . Zhang, J. Zhou, and J. Lu, “Gaussianformer: Scene as gaussians for vision-based 3d semantic occupancy predic- tion,”arXiv preprint arXiv:2405.17429, 2024

    2024 arXiv

  13. [21]

    Wild-gs: Real-time novel view synthesis from unconstrained photo collections,

    J. Xu, Y . Mei, and V . M. Patel, “Wild-gs: Real-time novel view synthesis from unconstrained photo collections,”arXiv preprint arXiv:2406.10373, 2024

    2024 arXiv

  14. [22]

    Gaussian in the wild: 3d gaussian splatting for unconstrained image collections,

    D. Zhang, C. Wang, W. Wang, P. Li, M. Qin, and H. Wang, “Gaussian in the wild: 3d gaussian splatting for unconstrained image collections,”arXiv preprint arXiv:2403.15704, 2024

    2024 arXiv

  15. [23]

    Wildgaussians: 3d gaussian splatting in the wild,

    J. Kulhanek, S. Peng, Z. Kukelova, M. Pollefeys, and T. Sattler, “Wildgaussians: 3d gaussian splatting in the wild,”arXiv preprint arXiv:2407.08447, 2024

    2024 arXiv

  16. [24]

    Splatfacto-w: A nerfstudio imple- mentation of gaussian splatting for unconstrained photo collections,

    C. Xu, J. Kerr, and A. Kanazawa, “Splatfacto-w: A nerfstudio imple- mentation of gaussian splatting for unconstrained photo collections,” arXiv preprint arXiv:2407.12306, 2024

    2024 arXiv

  17. [25]

    Optical models for direct volume rendering,

    N. Max, “Optical models for direct volume rendering,”IEEE Trans- actions on Visualization and Computer Graphics, vol. 1, no. 2, pp. 99–108, 1995

    1995

  18. [26]

    V olume rendering digest (for nerf),

    A. Tagliasacchi and B. Mildenhall, “V olume rendering digest (for nerf),”arXiv preprint arXiv:2209.02417, 2022

    2022 arXiv

  19. [27]

    Swag: Splatting in the wild images with appearance-conditioned gaussians,

    H. Dahmani, M. Bennehar, N. Piasco, L. Roldao, and D. Tsishkou, “Swag: Splatting in the wild images with appearance-conditioned gaussians,” 2024

    2024

  20. [28]

    Searching for mobilenetv3,

    A. Howard, M. Sandler, G. Chu, L.-C. Chen, B. Chen, M. Tan, W. Wang, Y . Zhu, R. Pang, V . Vasudevan,et al., “Searching for mobilenetv3,” inProceedings of the IEEE/CVF international conference on computer vision, pp. 1314–1324, 2019

    2019

  21. [29]

    Microsoft COCO: common objects in context,

    T. Lin, M. Maire, S. J. Belongie, L. D. Bourdev, R. B. Gir- shick, J. Hays, P. Perona, D. Ramanan, P. Doll ´ar, and C. L. Zitnick, “Microsoft COCO: common objects in context,”CoRR, vol. abs/1405.0312, 2014

    2014 arXiv

  22. [30]

    Spherical fibonacci mapping,

    B. Keinert, M. Innmann, M. S ¨anger, and M. Stamminger, “Spherical fibonacci mapping,”ACM Trans. Graph., vol. 34, nov 2015

    2015

  23. [31]

    Rethinking visual geo- localization for large-scale applications,

    G. Berton, C. Masone, and B. Caputo, “Rethinking visual geo- localization for large-scale applications,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 4878–4888, June 2022

    2022

  24. [32]

    Gaussian Splatting SLAM,

    H. Matsuki, R. Murai, P. H. J. Kelly, and A. J. Davison, “Gaussian Splatting SLAM,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024

    2024

  25. [33]

    Adam: A method for stochastic optimization,

    D. P. Kingma, “Adam: A method for stochastic optimization,”arXiv preprint arXiv:1412.6980, 2014

    2014 arXiv

  26. [34]

    Coco dataset stuff segmentation challenge,

    D. Puri, “Coco dataset stuff segmentation challenge,” in2019 5th international conference on computing, communication, control and automation (ICCUBEA), pp. 1–5, IEEE, 2019

    2019

  27. [35]

    Photo tourism: exploring photo collections in 3d,

    N. Snavely, S. M. Seitz, and R. Szeliski, “Photo tourism: exploring photo collections in 3d,” inACM siggraph 2006 papers, pp. 835– 846, 2006

    2006

  28. [36]

    Image quality assessment: from error visibility to structural similarity,

    Z. 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, vol. 13, no. 4, pp. 600–612, 2004

    2004

  29. [37]

    The unreasonable effectiveness of deep features as a perceptual metric,

    R. Zhang, P. Isola, A. A. Efros, E. Shechtman, and O. Wang, “The unreasonable effectiveness of deep features as a perceptual metric,” inProceedings of the IEEE conference on computer vision and pattern recognition, pp. 586–595, 2018

    2018

  30. [38]

    Colmap: A memory-efficient occupancy grid mapping framework,

    A. Fisher, R. Cannizzaro, M. Cochrane, C. Nagahawatte, and J. L. Palmer, “Colmap: A memory-efficient occupancy grid mapping framework,”Robotics and Autonomous Systems, vol. 142, p. 103755, 2021

    2021

Pith tools

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