Pith. sign in

REVIEW 4 major objections 4 minor 50 references

GSsplat: Generalizable Semantic Gaussian Splatting for Novel-view Synthesis in 3D Scenes

T0 review · 4 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A single feed-forward pass builds color and semantic Gaussian fields for unseen scenes, cutting inference time to under half a second.

desk verdict First generalizable semantic Gaussian splatting with joint color and semantics; real speed gains, but the SOTA claim is only tested against a subset of the cited baselines. read the letter →

arxiv 2505.04659 v1 pith:S5H25LJG submitted 2025-05-07 cs.GR

Feng Xiao , Hongbin Xu , Wanlin Liang , Wenxiong Kang This is my paper · ORCID
classification cs.GR
keywords generalizable3DGaussianSplattingnovel-viewsynthesissemanticsegmentationradiancefieldmulti-tasksceneunderstandingoffsetlearningdepth-guidedinitializationpoint-levelfeatureinteraction
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 3D Gaussian Splatting can be made generalizable and semantic at once, so that a model trained once on many scenes can synthesize novel-view colors and category maps for a scene it has never seen, without per-scene optimization. Prior generalizable semantic radiance fields are built on NeRF, which is slow because of dense ray sampling, whereas semantic Gaussian Splatting methods so far require scene-specific training. GSsplat takes K posed RGB views and depth priors, predicts scene-adaptive Gaussian positions and attributes in one forward pass, and renders both outputs with the same splatting operation. The reported upshot is state-of-the-art semantic synthesis on the ScanNet and Replica benchmarks at the fastest inference speed among comparable multi-task methods: 60.38 mIoU on ScanNet with eight views and estimated depth, in 0.48 seconds versus 9.62 seconds for the NeRF-based baseline.

What carries the argument

The central objects are 3D Gaussian primitives $G(x)=\exp(-\frac{1}{2}(x-\mu)^T\Sigma^{-1}(x-\mu))$ with covariance $\Sigma=RS S^T R^T$, together with per-Gaussian opacity, color coefficients, and semantic logits. The method's load-bearing mechanism is threefold: a hybrid multi-view encoder in which the color and semantic branches share the shallow convolutional blocks and split the deeper blocks; a point-level interaction module that treats the depth-unprojected pixels as a point cloud, partitions it into fixed-interval 3D units, and augments each point feature with an average-pooled, distance-weighted aggregate of its unit; and an offset-learning module that predicts an offset probability $\hat{t}_i$ and a translation $t_i$, moving only the centers whose probability exceeds 0.5 via $\mu_i = x_i + t_i \times \mathrm{Mask}(\hat{t}_i)$. The unmoved centers are pinned to the surface by the projective-depth consistency loss of Eq. (3). This machinery replaces the densification and pruning loops that prevent ordinary 3D Gaussian Splatting from generalizing to unseen scenes.

What would settle it

Run the trained model on the ScanNet evaluation set with deliberately corrupted depth maps—Gaussian noise of increasing standard deviation, or random dropout of depth regions—and measure mIoU and PSNR at each noise level. If semantic quality collapses before the offset module can compensate, the claim that offsets replace densification and pruning is falsified; if quality degrades gracefully, the depth-prior assumption is confirmed as the right characterization. The paper's own ablation removing offset supervision drops mIoU from 55.14 to 52.58 on Replica, so the supervision itself is measurably load-bearing.

Watch

Extended reading notes

Core claim

The central claim is that 3D Gaussian Splatting can be made both generalizable and semantically aware by replacing per-scene densification and pruning with two learned geometric mechanisms. Depth priors initialize Gaussian centers by unprojecting pixels; the paper's point-level interaction module aggregates neighboring-point features in local 3D units so each Gaussian's attributes reflect cross-view geometry, and the offset learning module splits centers into surface-staying and offset-moving groups, moving many of them into object-adjacent space to fit the scene. Offset-free centers are explicitly supervised by projective depth consistency, which keeps the geometry anchored. With ground-truth depth on ScanNet the method reaches 64.35 mIoU; with estimated depth from a cost-volume network it reaches 60.38 mIoU, outperforming the compared NeRF-based generalizable semantic field (58.30 mIoU) and the compared generalizable color-only Gaussian method (47.56 mIoU), while its total reconstruction-and-rendering time is 0.48 seconds versus 9.62 seconds. The authors conclude that generalizable Gaussian splatting with group-supervised offsets is the fastest path to semantic synthesis in unseen indoor scenes.

Load-bearing premise

The load-bearing premise is that the depth priors used to initialize and anchor the Gaussians are accurate enough; if the depth maps are wrong, the point cloud, the local 3D interaction neighborhoods, and the offset grouping are all built on incorrect geometry, and the paper gives no analysis of how much depth noise the offset module can absorb.

Editorial extensions

If this is right

  • Per-scene optimization disappears for semantic view synthesis: one forward pass over $K$ posed images and depth maps yields a renderable color-and-semantic Gaussian field for an unseen scene.
  • Inference becomes practical for interactive use: reconstructing and rendering a novel view takes 0.48 s with estimated depth and 0.22 s with ground-truth depth, compared with 9.62 s for the NeRF-based semantic field.
  • Accurate geometry is not a strict precondition for good semantics: with depth from a cost-volume network the mIoU gap to the ground-truth-depth setting is 3.97 points, not a collapse, so offsets compensate partly for noisy geometry.
  • The method remains usable with sparse input: at $K=4$ and $K=2$ on ScanNet the semantic mIoU stays at 59.51 and 57.57, well above the NeRF baseline at the same few views.
  • Separate color and semantic decoders, trained jointly, help both tasks: joint training raises semantic mIoU over an independently trained semantic model while preserving color quality.

Reading between the lines

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

  • Because the offset grouping relies on a fixed 0.5 probability threshold on predicted offsets, a stress test that perturbs depth maps with increasing noise would reveal how much of the claimed robustness comes from offsets versus the depth prior; the paper does not report such a sweep.
  • The same hybrid encoder and point-level interaction could attach other per-point attributes to splatted Gaussians—language embeddings, object instances, or uncertainty—making the architecture a general generalizable feature-splatting backbone rather than a semantic segmenter only.
  • The projective-depth supervision on unmoved centers could in principle be replaced by multi-view photometric consistency, which would remove the need for ground-truth depth during training and extend the method to scenes without depth sensors.
  • A stage-wise timing breakdown such as the paper's Table III suggests the marginal cost of adding a third Gaussian field is small because encoding and depth estimation dominate; adding more output tasks would therefore be cheap.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The manuscript proposes GSsplat, a generalizable 3D Gaussian Splatting method for joint color and semantic novel-view synthesis. Given K source views and depth maps (ground truth or CasMVSNet predictions), a hybrid multi-view encoder extracts color and semantic features, which are decoded to pixel-level feature vectors, unprojected to initialize Gaussian centers, refined by a point-level interaction module, and augmented with a grouped Gaussian offset learning scheme. The resulting color and semantic Gaussian radiance fields are rendered for novel views. Experiments on ScanNet and Replica compare GSsplat with GSNeRF, S-Ray, Neuray*, GeoNeRF*, and MVsplat*, reporting higher semantic mIoU and lower inference time than the evaluated baselines, together with ablations of the encoder structure, self-attention depth, offset learning, point-level interaction, and joint training.

Significance. Generalizable semantic radiance fields are currently dominated by NeRF-based methods that are slow, so adapting 3DGS to this setting with one-forward reconstruction and semantic rendering is a timely and potentially useful direction. The paper contributes a concrete architecture, a hybrid multi-task encoder, and two named modules (point-level interaction and offset learning with grouped supervision), and the ablations in Table VI support the usefulness of these components. The stage-wise timing in Table III is also informative. If the reported ScanNet result of 60.38 mIoU at 0.48 s for K=8 is reproducible and holds against all directly related baselines, this would be a meaningful advance in semantic view synthesis speed. However, the significance of the headline claim is currently qualified by incomplete baseline coverage and by missing statistical and evaluation details.

major comments (4)
  1. [Sec. IV-C, Tables I and II] The central state-of-the-art claim in the Abstract and Sec. IV-C is not fully supported because the two most directly related generalizable methods identified in the Related Work are not benchmarked. GNeSF [10] is a generalizable neural semantic field, and LatentSplat [40] is a generalizable 3DGS method that explicitly models semantic feature distributions, yet neither appears in Tables I or II. A single additional result above 60.38 mIoU, or a comparably accurate method faster than 0.48 s, would invalidate the claim as stated. The authors should either add these comparisons or restrict the claim to the specific set of evaluated methods.
  2. [Table I, K=4 and K=2 rows] The parenthetical values in the GSsplat rows for K=4 and K=2 (e.g., 6.09, 2.33, 1.51) are not defined in the caption or in the text, so the reduced-view robustness discussion in Sec. IV-C.4 cannot be fully interpreted. In addition, all metrics in Tables I and II are reported as single runs with no confidence intervals or repeated trials, which makes it difficult to judge whether margins such as the 2.08 mIoU advantage over GSNeRF on ScanNet K=8 are meaningful.
  3. [Sec. III-C.2, Eq. (2)] The paper does not specify how gradients flow through the hard threshold Mask(·) that selects Gaussian centers with offset probability above 0.5. If Mask is a binary, non-differentiable operator, then the offset probability t̂ receives no gradient from either the rendering loss or the offset supervision in Eq. (3), and it is unclear how the offset predictor is trained. The authors should describe the differentiable relaxation, straight-through estimator, or alternative mechanism used for this selection.
  4. [Sec. III-C and Eq. (3)] The initialization of Gaussian centers and the offset supervision in Eq. (3) both rely on the same multi-view depth maps, yet the paper does not analyze how depth estimation errors propagate. Unprojection of inaccurate MVS depth determines the point cloud, the point-level interaction, and the offset grouping, so the central premise is that the depth priors are sufficiently accurate. An experiment with perturbed or noisy depth, or a discussion of failure cases under the mvs setting, is needed to support the robustness implied by the method description.
minor comments (4)
  1. [Sec. III-C.1, Eq. (1)] The notation is imprecise: μ0 ∈ R^N should be μ0 ∈ R^(N×3), and 'V_i(i ∈ N)' should read 'V_i (i = 1, ..., N)' to avoid confusing the index set with the point-count dimension.
  2. [Sec. V, Conclusion] The phrase 'fastest free-forward speed' should be corrected to 'fastest feed-forward speed'.
  3. [Tables I and II, star markers] The starred baselines (Neuray*, GeoNeRF*, MVsplat*) are described as having 'an extra semantic rendering module' incorporated, but the implementation of that module for each baseline is not described; a brief appendix entry or a reference to an existing implementation would be needed for reproducibility.
  4. [Table I, caption] The statement that the K=8 NeRF-based results are derived from [11] should be accompanied by a confirmation that the dataset splits, preprocessing, and evaluation metrics are identical to those used for GSsplat; otherwise the comparison may inherit differences in evaluation protocol.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the central result is an empirical comparison on held-out benchmarks with standard supervised training.

full rationale

GSsplat's central claim is an empirical accuracy/speed comparison on held-out ScanNet and Replica scenes. Training uses standard supervised losses (Eqs. 5-7): RGB MSE/LPIPS against target images, cross-entropy against semantic labels, depth loss against ground-truth depths for the MVSNet depth estimator, and the geometric invariance loss Lf. None of these losses is used to fit the reported test metrics. The offset loss Lf (Eq. 3) is the only self-referential component: non-offset Gaussian centers are initialized by unprojecting source depth maps and then supervised to project back to those same depths across views. This is a geometric consistency regularizer rather than a fitted parameter renamed as a prediction; it does not determine novel-view semantic labels. The method is benchmarked against external methods (GSNeRF, S-Ray, Neuray, GeoNeRF, MVsplat) and test scenes are unseen. No load-bearing uniqueness theorem or self-citation is invoked; the cited prior work supplies architectures and baselines rather than the paper's conclusions. Therefore no circular step can be exhibited by quotation and reduction. Concerns about missing baselines such as GNeSF or LatentSplat bear on comparison completeness, not on whether the derivation reduces to its inputs by construction.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The method reuses standard 3DGS math and relies on depth priors and 2D semantic supervision rather than postulating new physical entities. The main load-bearing assumptions are the quality of the depth priors and the validity of the depth-invariance regularizer.

free parameters (4)
  • lambda1 (color MSE loss weight) = 10.0
    Set in Section IV-B and used in Eq. 5; chosen by hand, affects color reconstruction loss balance.
  • lambda_f (offset geometric invariance loss weight) = 0.2
    Set in Section IV-B and used in Eq. 7; chosen by hand, controls strength of offset supervision.
  • offset mask threshold = 0.5
    Used in Eq. 2 to split Gaussian centers into offset and non-offset groups; a hand-set prior threshold.
  • number of self-attention layers in semantic encoder = 3
    Selected by ablation in Table V; architecture hyperparameter tuned on validation.
assumptions (4)
  • standard math 3D Gaussian Splatting represents a scene as a set of learnable 3D Gaussians and renders via differentiable alpha-blending (Kerbl et al., 3DGS).
    The paper adopts the 3DGS rendering formulation in Section III-A without formal proof.
  • domain assumption Multi-view depth priors (ground-truth or CasMVSNet estimates) are sufficiently accurate to initialize Gaussian centers.
    Section III-C unprojects pixel features using predicted or GT depths; the training pipeline first trains the depth network, and the whole construction depends on these depths.
  • domain assumption The non-offset Gaussian centers should satisfy projective depth invariance with the source-view depth maps.
    Used to define Lf in Eq. 3; assumes the depth maps are consistent and that surface points project back to the same depth.
  • domain assumption 2D semantic labels on source views provide sufficient supervision for 3D semantic consistency.
    Semantic loss in Eq. 6 supervises both rendered and source-view semantic maps with 2D labels; cross-view consistency is assumed to be learned implicitly.

how reviews work

0 comments
Cite this review

Pith. "Pith review of GSsplat: Generalizable Semantic Gaussian Splatting for Novel-view Synthesis in 3D Scenes." pith.science (2026). https://pith.science/paper/S5H25LJG

@misc{pith2026250504659,
  author       = {Pith},
  title        = {Pith review of: GSsplat: Generalizable Semantic Gaussian Splatting for Novel-view Synthesis in 3D Scenes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/S5H25LJG}},
  note         = {Machine review of arXiv:2505.04659}
}
read the original abstract

The semantic synthesis of unseen scenes from multiple viewpoints is crucial for research in 3D scene understanding. Current methods are capable of rendering novel-view images and semantic maps by reconstructing generalizable Neural Radiance Fields. However, they often suffer from limitations in speed and segmentation performance. We propose a generalizable semantic Gaussian Splatting method (GSsplat) for efficient novel-view synthesis. Our model predicts the positions and attributes of scene-adaptive Gaussian distributions from once input, replacing the densification and pruning processes of traditional scene-specific Gaussian Splatting. In the multi-task framework, a hybrid network is designed to extract color and semantic information and predict Gaussian parameters. To augment the spatial perception of Gaussians for high-quality rendering, we put forward a novel offset learning module through group-based supervision and a point-level interaction module with spatial unit aggregation. When evaluated with varying numbers of multi-view inputs, GSsplat achieves state-of-the-art performance for semantic synthesis at the fastest speed.

Figures

Figures reproduced from arXiv: 2505.04659 by the authors.

Figure 1
Figure 1. The top is the non-generalizable 3DGS reconstruction process and [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of GSsplat. Given N source views and camera poses, our model directly predicts the semantic Gaussian and color Gaussian parameters from RGB and depth information for 3D scene reconstruction. Firstly, the hybrid network uses a multi-view encoding module to extract 2D semantic and color features. Next, the features are decoded to the original image resolution and unprojected per pixel to the 3D space. After p… view at source ↗
Figure 4
Figure 4. Offset learning module. We divide the initial Gaussian centers (red) into two groups, one with offset (green) and one without offset (blue), and supervise them from both the rendering and geometric projection directions. Gaussian splitting rendering are determined by the contribution of each Gaussian distribution in the observation direction, while for dense view reconstruction, the contributions of Gaus￾sian points… view at source ↗
Figures from the paper (4 more)
Figure 3
Figure 3. Figure 3: Point-level interaction module. The left is the cross-view interaction method based on 2D features in other multi-view reconstruction approaches, and the right presents the process of our point-level interaction through space unit aggregation [PITH_FULL_IMAGE:figures/…
Figure 5
Figure 5. Figure 5: The demonstration of the novel-view synthesis. Our method is compared with GSNeRF [11] on 8-view inputs on evaluation datasets. The first two rows are from the test data of ScanNet, and the last two rows are from Replica [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 7
Figure 7. Figure 7: The structure of our proposed multi-view network with different [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Results of 2 novel-views of one of the Replica test scenes conducted during the ablation studies. The first column in each block is the rendered color [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

50 extracted references · 20 canonical work pages

  1. [10]

    Gnesf: Generalizable neural semantic fields,

    H. Chen, C. Li, M. Guo, Z. Yan, and G. H. Lee, “Gnesf: Generalizable neural semantic fields,” Advances in Neural Information Processing Systems, vol. 36, pp. 36 553–36 565, 2023

    work page 2023

  2. [40]

    latentsplat: Autoencoding variational gaussians for fast generalizable 3d reconstruc- tion,

    C. Wewer, K. Raj, E. Ilg, B. Schiele, and J. E. Lenssen, “latentsplat: Autoencoding variational gaussians for fast generalizable 3d reconstruc- tion,” arXiv preprint arXiv:2403.16292 , 2024

    arXiv 2024

  3. [1]

    Recent advances in multi-modal 3d scene understanding: A comprehensive survey and evaluation,

    Y . Lei, Z. Wang, F. Chen, G. Wang, P. Wang, and Y . Yang, “Recent advances in multi-modal 3d scene understanding: A comprehensive survey and evaluation,” arXiv preprint arXiv:2310.15676 , 2023

    arXiv 2023

  4. [2]

    Deep learning based 3d segmentation: A survey,

    Y . He, H. Yu, X. Liu, Z. Yang, W. Sun, and A. Mian, “Deep learning based 3d segmentation: A survey,” arXiv preprint arXiv:2103.05423 , 2021

    arXiv 2021

  5. [3]

    Semantically-aware neural radiance fields for vi- sual scene understanding: A comprehensive review,

    T.-A.-Q. Nguyen, A. Bourki, M. Macudzinski, A. Brunel, and M. Bennamoun, “Semantically-aware neural radiance fields for vi- sual scene understanding: A comprehensive review,” arXiv preprint arXiv:2402.11141, 2024

    arXiv 2024

  6. [4]

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

    B. Mildenhall, P. P. Srinivasan, M. Tancik, J. T. Barron, R. Ramamoorthi, and R. Ng, “Nerf: Representing scenes as neural radiance fields for view synthesis,” Communications of the ACM , vol. 65, no. 1, pp. 99–106, 2021

    2021

  7. [5]

    In-place scene labelling and understanding with implicit scene representation,

    S. Zhi, T. Laidlow, S. Leutenegger, and A. J. Davison, “In-place scene labelling and understanding with implicit scene representation,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 15 838–15 847

    2021

  8. [6]

    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 Trans. Graph. , vol. 42, no. 4, pp. 139–1, 2023

    2023

Show all 50 references
  1. [7]

    A survey on 3d gaussian splatting,

    G. Chen and W. Wang, “A survey on 3d gaussian splatting,” arXiv preprint arXiv:2401.03890, 2024

    2024 arXiv

  2. [8]

    Semantic gaussians: Open-vocabulary scene understanding with 3d gaussian splatting,

    J. Guo, X. Ma, Y . Fan, H. Liu, and Q. Li, “Semantic gaussians: Open-vocabulary scene understanding with 3d gaussian splatting,” arXiv preprint arXiv:2403.15624, 2024

    2024 arXiv

  3. [9]

    Semantic ray: Learning a generalizable semantic field with cross-reprojection attention,

    F. Liu, C. Zhang, Y . Zheng, and Y . Duan, “Semantic ray: Learning a generalizable semantic field with cross-reprojection attention,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 17 386–17 396

    2023

  4. [11]

    Gsnerf: Gen- eralizable semantic neural radiance fields with enhanced 3d scene un- derstanding,

    Z.-T. Chou, S.-Y . Huang, I. Liu, Y .-C. F. Wang et al. , “Gsnerf: Gen- eralizable semantic neural radiance fields with enhanced 3d scene un- derstanding,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 20 806–20 815. 10

    2024

  5. [12]

    Rt-gs2: Real-time generalizable semantic segmentation for 3d gaussian representations of radiance fields,

    M.-B. Jurca, R. Royen, I. Giosan, and A. Munteanu, “Rt-gs2: Real-time generalizable semantic segmentation for 3d gaussian representations of radiance fields,” arXiv preprint arXiv:2405.18033 , 2024

    2024 arXiv

  6. [13]

    Mvsplat: Efficient 3d gaussian splatting from sparse multi-view images,

    Y . Chen, H. Xu, C. Zheng, B. Zhuang, M. Pollefeys, A. Geiger, T.- J. Cham, and J. Cai, “Mvsplat: Efficient 3d gaussian splatting from sparse multi-view images,” in European Conference on Computer Vision. Springer, 2025, pp. 370–386

    2025

  7. [14]

    Georgs: Geomet- ric regularization for real-time novel view synthesis from sparse inputs,

    Z. Liu, J. Su, G. Cai, Y . Chen, B. Zeng, and Z. Wang, “Georgs: Geomet- ric regularization for real-time novel view synthesis from sparse inputs,” IEEE Transactions on Circuits and Systems for Video Technology , 2024

    2024

  8. [15]

    Depth- guided robust point cloud fusion nerf for sparse input views,

    S. Guo, Q. Wang, Y . Gao, R. Xie, L. Li, F. Zhu, and L. Song, “Depth- guided robust point cloud fusion nerf for sparse input views,” IEEE Transactions on Circuits and Systems for Video Technology , 2024

    2024

  9. [16]

    Fsgs: Real-time few-shot view synthesis using gaussian splatting,

    Z. Zhu, Z. Fan, Y . Jiang, and Z. Wang, “Fsgs: Real-time few-shot view synthesis using gaussian splatting,” in European conference on computer vision. Springer, 2025, pp. 145–163

    2025

  10. [17]

    A multi-task network for multi- view stereo reconstruction: When semantic consistency based clustering meets depth estimation optimization,

    X. Huang, S. Zhang, J. Li, and L. Wang, “A multi-task network for multi- view stereo reconstruction: When semantic consistency based clustering meets depth estimation optimization,” IEEE Transactions on Geoscience and Remote Sensing , 2024

    2024

  11. [18]

    Nesf: Neural semantic fields for generalizable semantic segmentation of 3d scenes,

    S. V ora, N. Radwan, K. Greff, H. Meyer, K. Genova, M. S. Sajjadi, E. Pot, A. Tagliasacchi, and D. Duckworth, “Nesf: Neural semantic fields for generalizable semantic segmentation of 3d scenes,” arXiv preprint arXiv:2111.13260, 2021

    2021 arXiv

  12. [19]

    Panoptic nerf: 3d-to-2d label transfer for panoptic urban scene segmentation,

    X. Fu, S. Zhang, T. Chen, Y . Lu, L. Zhu, X. Zhou, A. Geiger, and Y . Liao, “Panoptic nerf: 3d-to-2d label transfer for panoptic urban scene segmentation,” in 2022 International Conference on 3D Vision (3DV) . IEEE, 2022, pp. 1–11

    2022

  13. [20]

    Or-nerf: Object removing from 3d scenes guided by multiview segmentation with neural radiance fields,

    Y . Yin, Z. Fu, F. Yang, and G. Lin, “Or-nerf: Object removing from 3d scenes guided by multiview segmentation with neural radiance fields,” arXiv preprint arXiv:2305.10503 , 2023

    2023 arXiv

  14. [21]

    Lerf: Language embedded radiance fields,

    J. Kerr, C. M. Kim, K. Goldberg, A. Kanazawa, and M. Tancik, “Lerf: Language embedded radiance fields,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2023, pp. 19 729–19 739

    2023

  15. [22]

    Open-nerf: Towards open vocabulary nerf decomposition,

    H. Zhang, F. Li, and N. Ahuja, “Open-nerf: Towards open vocabulary nerf decomposition,” in Proceedings of the IEEE/CVF Winter Confer- ence on Applications of Computer Vision , 2024, pp. 3456–3465

    2024

  16. [23]

    Gov-nesf: Generalizable open- vocabulary neural semantic fields,

    Y . Wang, H. Chen, and G. H. Lee, “Gov-nesf: Generalizable open- vocabulary neural semantic fields,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 20 443–20 453

    2024

  17. [24]

    Ov-nerf: Open-vocabulary neural radiance fields with vision and language foundation models for 3d semantic understanding,

    G. Liao, K. Zhou, Z. Bao, K. Liu, and Q. Li, “Ov-nerf: Open-vocabulary neural radiance fields with vision and language foundation models for 3d semantic understanding,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 34, no. 12, pp. 12 923–12 936, 2024

    2024

  18. [25]

    Recent advances in 3d gaussian splatting,

    T. Wu, Y .-J. Yuan, L.-X. Zhang, J. Yang, Y .-P. Cao, L.-Q. Yan, and L. Gao, “Recent advances in 3d gaussian splatting,” Computational Visual Media, vol. 10, no. 4, pp. 613–642, 2024

    2024

  19. [26]

    Gaussian grouping: Segment and edit anything in 3d scenes,

    M. Ye, M. Danelljan, F. Yu, and L. Ke, “Gaussian grouping: Segment and edit anything in 3d scenes,” arXiv preprint arXiv:2312.00732, 2023

    2023 arXiv

  20. [27]

    2d-guided 3d gaussian segmentation,

    K. Lan, H. Li, H. Shi, W. Wu, Y . Liao, L. Wang, and P. Zhou, “2d-guided 3d gaussian segmentation,” arXiv preprint arXiv:2312.16047 , 2023

    2023 arXiv

  21. [28]

    Language embedded 3d gaussians for open-vocabulary scene understanding,

    J.-C. Shi, M. Wang, H.-B. Duan, and S.-H. Guan, “Language embedded 3d gaussians for open-vocabulary scene understanding,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion, 2024, pp. 5333–5343

    2024

  22. [29]

    Feature 3dgs: Supercharging 3d gaussian splatting to enable distilled feature fields,

    S. Zhou, H. Chang, S. Jiang, Z. Fan, Z. Zhu, D. Xu, P. Chari, S. You, Z. Wang, and A. Kadambi, “Feature 3dgs: Supercharging 3d gaussian splatting to enable distilled feature fields,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, p...

    2024

  23. [30]

    Langsplat: 3d language gaussian splatting,

    M. Qin, W. Li, J. Zhou, H. Wang, and H. Pfister, “Langsplat: 3d language gaussian splatting,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 20 051–20 060

    2024

  24. [31]

    Segment anything,

    A. Kirillov, E. Mintun, N. Ravi, H. Mao, C. Rolland, L. Gustafson, T. Xiao, S. Whitehead, A. C. Berg, W.-Y . Loet al., “Segment anything,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 4015–4026

    2023

  25. [32]

    Opengaussian: Towards point- level 3d gaussian-based open vocabulary understanding,

    Y . Wu, J. Meng, H. Li, C. Wu, Y . Shi, X. Cheng, C. Zhao, H. Feng, E. Ding, J. Wang et al. , “Opengaussian: Towards point- level 3d gaussian-based open vocabulary understanding,” arXiv preprint arXiv:2406.02058, 2024

    2024 arXiv

  26. [33]

    3d gaussian splatting: Survey, technologies, challenges, and opportunities,

    Y . Bao, T. Ding, J. Huo, Y . Liu, Y . Li, W. Li, Y . Gao, and J. Luo, “3d gaussian splatting: Survey, technologies, challenges, and opportunities,” arXiv preprint arXiv:2407.17418 , 2024

    2024 arXiv

  27. [34]

    pixelnerf: Neural radiance fields from one or few images,

    A. Yu, V . Ye, M. Tancik, and A. Kanazawa, “pixelnerf: Neural radiance fields from one or few images,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2021, pp. 4578– 4587

    2021

  28. [35]

    Featurenerf: Learning generalizable nerfs by distilling foundation models,

    J. Ye, N. Wang, and X. Wang, “Featurenerf: Learning generalizable nerfs by distilling foundation models,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2023, pp. 8962–8973

    2023

  29. [36]

    Insertnerf: Instill- ing generalizability into nerf with hypernet modules,

    Y . Bao, T. Ding, J. Huo, W. Li, Y . Li, and Y . Gao, “Insertnerf: Instill- ing generalizability into nerf with hypernet modules,” arXiv preprint arXiv:2308.13897, 2023

    2023 arXiv

  30. [37]

    Hybrid-mvs: Robust multi-view reconstruction with hybrid optimization of visual and depth cues,

    C. Li, L. Zhou, H. Jiang, Z. Zhang, X. Xiang, H. Sun, Q. Luan, H. Bao, and G. Zhang, “Hybrid-mvs: Robust multi-view reconstruction with hybrid optimization of visual and depth cues,” IEEE Transactions on Circuits and Systems for Video Technology , vol. 33, no. 12, pp. 7630– 7644, 2023

    2023

  31. [38]

    Robustmvs: Single domain generalized deep multi-view stereo,

    H. Xu, W. Chen, B. Sun, X. Xie, and W. Kang, “Robustmvs: Single domain generalized deep multi-view stereo,” IEEE Transactions on Circuits and Systems for Video Technology , 2024

    2024

  32. [39]

    pixelsplat: 3d gaussian splats from image pairs for scalable generalizable 3d recon- struction,

    D. Charatan, S. L. Li, A. Tagliasacchi, and V . Sitzmann, “pixelsplat: 3d gaussian splats from image pairs for scalable generalizable 3d recon- struction,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 19 457–19 467

    2024

  33. [41]

    Gaussian splatting: 3d reconstruction and novel view synthesis, a review,

    A. Dalal, D. Hagen, K. G. Robbersmyr, and K. M. Knausg ˚ard, “Gaussian splatting: 3d reconstruction and novel view synthesis, a review,” IEEE Access, 2024

    2024

  34. [42]

    Flash3d: Feed-forward gener- alisable 3d scene reconstruction from a single image,

    S. Szymanowicz, E. Insafutdinov, C. Zheng, D. Campbell, J. F. Hen- riques, C. Rupprecht, and A. Vedaldi, “Flash3d: Feed-forward gener- alisable 3d scene reconstruction from a single image,” arXiv preprint arXiv:2406.04343, 2024

    2024 arXiv

  35. [43]

    Cascade cost volume for high-resolution multi-view stereo and stereo matching,

    X. Gu, Z. Fan, S. Zhu, Z. Dai, F. Tan, and P. Tan, “Cascade cost volume for high-resolution multi-view stereo and stereo matching,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 2495–2504

    2020

  36. [44]

    Neural rays for occlusion-aware image-based rendering,

    Y . Liu, S. Peng, L. Liu, Q. Wang, P. Wang, C. Theobalt, X. Zhou, and W. Wang, “Neural rays for occlusion-aware image-based rendering,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 7824–7833

    2022

  37. [45]

    Geonerf: Generalizing nerf with geometry priors,

    M. M. Johari, Y . Lepoittevin, and F. Fleuret, “Geonerf: Generalizing nerf with geometry priors,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 18 365–18 375

    2022

  38. [46]

    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, 2017, pp. 5828–5839

    2017

  39. [47]

    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

    1906 arXiv

  40. [48]

    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

  41. [49]

    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,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 586–595

    2018

  42. [50]

    Adam: A method for stochastic optimization,

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

    2014 arXiv

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.