Pith. sign in

REVIEW 4 major objections 6 minor 35 references

OutLangSplat: 3D Language Gaussian Splatting for UAV Outdoor Scenes

T0 review · 4 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read OutLangSplat adapts language Gaussian splatting to UAV outdoor scenes, with 88.8% localization and up to 82.9% mIoU.

desk verdict Useful new UAV dataset and a sensible method, but test-set hyperparameter tuning makes the SOTA claim conditional. read the letter →

arxiv 2608.04560 v1 pith:VHON34YC submitted 2026-08-05 cs.CV cs.GR

Xia Yan , He Wu , Yanghui Xu , Zizhao Wu , Jiazhou Chen This is my paper · ORCID
classification cs.CVcs.GR
keywords 3DGaussianSplattingopen-vocabularysceneunderstandingUAVoutdoorscenessemanticsegmentationinstancelocalizationmulti-viewfeatureaggregationremotesensingvision-languagemodelsdataset
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 tries to establish that 3D Language Gaussian Splatting, which embeds open-vocabulary text features into 3D Gaussians, can be made to work reliably in large outdoor scenes seen from low-altitude UAV cameras. The proposed system, OutLangSplat, combines 2D vision-language features with 3D structural features at the level of semantic regions, then aggregates multi-view observations onto Gaussians while weighting down pixels and views that are unreliable due to occlusion, boundary mixing, or cross-view disagreement. On a new manually annotated benchmark built from four public UAV scenes, the method reports 88.8% Loc@50 for instance localization and up to 82.9% mIoU for semantic segmentation. A sympathetic reader should care because open-vocabulary 3D understanding, querying a reconstructed scene with arbitrary text rather than a fixed label set, has so far been demonstrated mostly in indoor or small-scale settings, and this work provides both a method and a dataset aimed at geographic objects in UAV imagery.

What carries the argument

The machinery is a two-stage semantic representation pipeline built on 3D Gaussian Splatting. First, semantic regions from SAM serve as alignment anchors: 2D region features from RemoteCLIP and RS5M are fused with average-pooled 3D structural features produced by a SegContrast-initialized MinkUNet, mapped into the vision-language space by a two-layer MLP. Second, an iterative aggregation assigns each Gaussian a language feature by weighting pixel observations with an effective contribution weight that combines alpha-blending weight, a valid-mask factor, a sum-completeness clip, and an inverse-Simpson-index penalty for dispersed contributions, then reweights each view by exponential cosine consistency with the current aggregate. This reliability weighting is what suppresses occluded, boundary, and cross-view noise without discarding low-contribution surface Gaussians.

What would settle it

Take one of the four benchmark scenes, corrupt the camera poses or the reconstructed Gaussian positions (or use a low-quality reconstruction with visible artifacts), and run OutLangSplat; if localization accuracy and segmentation IoU do not drop measurably below the reported 88.8% Loc@50 and 82.9% mIoU, the paper's stated dependence on reconstruction quality would not be load-bearing.

Watch

Extended reading notes

Core claim

The central claim is that the failure of existing language-embedded 3DGS methods in UAV outdoor scenes comes from two fixable weaknesses: features that lack 3D spatial structure, and aggregation that treats all pixels and views as equally trustworthy. OutLangSplat addresses the first by extracting region-level features from two complementary vision-language encoders, projecting the 3D Gaussian point cloud into each view, and fusing pooled 3D structural features into those 2D semantic regions through a learned mapping. It addresses the second with a training-free aggregation that computes a per-pixel reliability from the dispersion of Gaussian alpha-blending contributions, measured by an inverse Simpson index, and per-view reliability from iterative cosine consistency between a view's observation and the aggregated feature. The paper reports that this combination outperforms LangSplat, LUDVIG, and Lang3D-XL on both semantic segmentation and instance localization over its four-scene UAV benchmark.

Load-bearing premise

The pipeline assumes the initial 3D Gaussian reconstruction is geometrically accurate enough that projecting 3D points into 2D views yields correct correspondences; if the reconstruction has artifacts, incomplete geometry, or drifting camera poses, the region alignment and pixel-to-Gaussian reliability weights lose their meaning.

Editorial extensions

If this is right

  • Open-vocabulary segmentation and localization can transfer from indoor scenes to large UAV outdoor scenes with the same 3DGS backbone, without retraining a semantic field.
  • Weighting multi-view feature aggregation by pixel contribution concentration and cross-view consistency improves mIoU and localization accuracy over naive projection aggregation and over visibility-gated aggregation.
  • Fusing 3D structural features at region level reduces incomplete target responses and background misactivations compared with using either 2D encoder alone or 2D fusion without 3D structure.
  • The released four-scene, 660-region benchmark provides a common evaluation basis for open-vocabulary UAV scene understanding, enabling future methods to be compared on semantic segmentation and instance localization together.
  • Because the aggregation module is training-free, the added cost over LUDVIG-like lifting is small, making the approach practical on a single NVIDIA 3090.

Reading between the lines

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

  • The reliability-weighting idea is not tied to 3DGS: the inverse-Simpson dispersion penalty and iterative consistency reweighting could be applied to any multi-view feature lifting scheme, including NeRF-based feature fields or 2D video segmentation.
  • Because the method inherits the reconstruction from an unmodified 3DGS run, its accuracy on small or heavily occluded objects is bounded by reconstruction completeness; joint optimization of geometry and language features is the natural next step.
  • The benchmark's manual polygon annotations across views could support a follow-up study on annotation cost versus performance, or be extended to temporal UAV sequences where the same object appears at very different scales.
  • The reported gains are on four scenes; whether the margin holds on a wider variety of UAV altitudes, seasons, and object categories is a testable open question.
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 / 6 minor

Summary. OutLangSplat proposes a 3D Gaussian Splatting language-feature framework for open-vocabulary semantic segmentation and instance localization in UAV outdoor scenes. It combines two region-level vision-language encoders (RemoteCLIP and RS5M) with a sparse-convolution 3D structural branch, distills the 3D features into the 2D feature space via a learned MLP, and then aggregates multi-view features onto Gaussians using pixel-level contribution reliability and an iterative cross-view consistency weighting. The authors introduce a manually annotated dataset of 660 regions across four public UAV scenes and report experiments comparing with LangSplat, LUDVIG, and Lang3D-XL, claiming state-of-the-art localization (88.8% Loc@50 overall) and segmentation (up to 82.9% mIoU). The paper also includes ablations of the feature fusion and aggregation components.

Significance. The proposed dataset addresses a real gap: UAV outdoor scenes are underrepresented in open-vocabulary 3D scene understanding benchmarks. The aggregation strategy is conceptually simple and training-free, which is appealing for practical use, and the paper provides quantitative ablations for its main components. The promise to open-source data and code is also a positive step. However, the confidence in the headline quantitative claims is limited by test-set hyperparameter selection, undocumented baseline tuning, and the absence of error bars or statistical tests. These issues are fixable within the manuscript's scope, but they currently make the claimed superiority conditional.

major comments (4)
  1. [Experiments, Table 5] The fusion weights λ2D and λf are selected by evaluating on the same annotations used in the headline comparisons in Tables 2–4, and no held-out validation split or selection protocol is described. Since Table 5 shows that adjacent settings change mIoU and mLoc@50 by several points (e.g., mLoc@50 ranges from 86.2 to 91.6 for λ2D=0.6), the final configuration appears to be the one that maximizes the reported metrics on the test scenes. This gives the method an oracle advantage over the baselines and inflates the reported margins. Please add a proper validation split or a separate parameter-selection procedure, and report the sensitivity of the Tables 2–3 numbers to the chosen weights.
  2. [Experiments, baselines] The comparison to the three baselines is not documented sufficiently. The paper does not state whether LangSplat, LUDVIG, and Lang3D-XL were run with their default hyperparameters, whether any per-scene tuning was performed for them, or whether they used the same 3DGS reconstructions and vision-language features. Because the SOTA claim rests on these comparisons, please report the exact settings, any tuning performed, and the computational budget for each method, and make the configuration files available.
  3. [Region-based 2D-3D feature fusion] The 3D structural branch is trained with the distillation loss L_distill = 1 - cos(h_theta(f_3D), f_2D), which regresses onto the same 2D semantic teacher used in the 2D branch. Therefore the 3D features do not introduce independent semantic evidence; any improvement could come from spatial smoothing or architectural regularization rather than from genuinely complementary 3D structure. To support the claim that 3D structural information improves semantic representation, please add a control experiment that separates the learned mapping from a purely geometric regularizer, for example by replacing the learned MLP with a fixed feature-propagation or smoothing baseline, or by training the 3D branch on a different task.
  4. [Experiments, evaluation protocol] No error bars or statistical significance measures are reported. Tables 2–3 present single-run numbers, and several differences are small (e.g., Table 2, Campus Loc@40 is 87.1 for both LangSplat and Ours). Given the manual annotation process and the small number of scenes, the reported margins may not be stable. Please report means and variances over repeated runs or confidence intervals, and provide annotation quality measures such as inter-annotator agreement.
minor comments (6)
  1. [Experimental Setup, Dataset] The description of the annotation process is thin: it mentions a self-developed script and about one month of work, but does not state the number of annotators, the annotation guidelines, or how ambiguous boundaries and occluded regions were resolved. Please add inter-annotator agreement statistics and a sample of the annotation interface.
  2. [Eq. (8)] The inverse Simpson index is written with an epsilon in the denominator; please clarify the value of epsilon and confirm that the weights π are normalized before applying the formula.
  3. [Table 5] The table is hard to read because the triples under mIoU, mAcc, and mLoc@50 correspond to λf values 0.6/0.7/0.8, but this is not stated in the header. Please make the table self-contained by listing λf explicitly.
  4. [Throughout] The capitalization of the Ludvig baseline is inconsistent ("Ludvig" in Table 3 versus "LUDVIG" elsewhere); please standardize it.
  5. [Experimental Setup] The text says "more results are in the supplementary material," but no supplementary material is provided with the arXiv submission. Please list what additional results are planned and make them available with the revision.
  6. [Conclusion and Contributions] The paper claims the "first accessible" open-vocabulary 3D dataset for UAV scenes. Please clarify how this claim was verified against existing datasets and whether the underlying public scenes already contain annotations that could be reused, to avoid overstating novelty.

Circularity Check

1 steps flagged · score 6.0 of 10

Headline SOTA margins are partly a test-set fit: λ2D and λf are tuned on the same scenes used for Tables 2-3.

  1. fitted input called prediction [Ablation on feature fusion, Table 5 (with Eqs. 5-7); final comparisons in Tables 2-3]
    "We further evaluate λ2D ∈ {0.5,0.6,0.7} and λf ∈ {0.6,0.7,0.8}. As shown in Table 5, the best overall performance is obtained when λ2D = 0.6 and λf = 0.7."

    The fusion weights λ2D and λf are selected by maximizing mIoU/mAcc/mLoc@50 in Table 5 on the same annotated scenes that are then used for the headline comparisons in Tables 2-3. No validation/test split is described. Adjacent grid settings shift the metrics by several points (e.g., at λ2D=0.6, mLoc@50 ranges from 86.2 at λf=0.8 to 91.6 at λf=0.7), so the reported Loc@50=88.8% and mIoU up to 82.9% are partly a test-set selection result rather than an independent prediction. The baselines are given no analogous oracle tuning, so the SOTA margins in Tables 2-3 are statistically forced and the comparison is conditional on the chosen configuration.

full rationale

The main comparative evaluation is against manually annotated ground truth, which is external to the method's fitted values, so the benchmark claim is not inherently circular. The 2D-3D fusion uses an explicit distillation loss to map 3D features into the 2D semantic space; this is a design choice that does not by itself make the evaluation circular, though it limits the independence of the 3D semantic content. The stated limitation in 'Limitations and Future Works' (dependency on 3DGS reconstruction quality) is acknowledged and applies to baselines as well, so it is not a circularity. References to prior work (3DGS, LangSplat, SAM, etc.) are external and not load-bearing self-citations. The only concrete circularity is the test-set selection of the two fusion hyperparameters, which directly shapes the headline numbers. This makes the SOTA claim partially circular: the reported margins are partly a product of selecting the best configuration on the evaluation data, while the architecture, dataset, and aggregation strategy still carry independent content.

Assumptions & free parameters 11 free parameters · 5 assumptions · 0 invented entities

The method depends on several domain assumptions about the reliability of reconstructed geometry, SAM region quality, and the complementarity of the two vision-language encoders. The 3D structural branch adds no independent semantic information because it is distilled to match the same 2D semantic features, so the semantic power ultimately comes from the pretrained 2D models. Many hyperparameters are set by hand or tuned on the evaluation dataset.

free parameters (11)
  • lambda_2D = 0.6
    Fusion weight between RemoteCLIP and RS5M region features; chosen by ablation on the test dataset (Table 5).
  • lambda_f = 0.7
    Fusion weight between 2D semantic feature and mapped 3D structural feature; chosen by ablation on the test dataset (Table 5).
  • tau_E = 12
    Threshold on effective number of contributing Gaussians in Eq. 8; set by hand in all experiments.
  • kappa = 8
    Decay strength in q_eff; set by hand in all experiments.
  • tau_d
    Visibility threshold for retaining view-level observations (O_i); value not specified in the paper.
  • tau_o
    Opacity threshold for filtering unreliable Gaussians before downsampling; value not specified.
  • voxel_size_s
    Voxel size in the sparse convolutional input; value not specified.
  • beta_t
    Sharpening coefficient in the consistency-aware iteration (Eq. 12); schedule not specified.
  • T
    Number of consistency aggregation iterations; not specified.
  • q_min
    Lower clip bound in q_sum; not specified.
  • topk_coverage_threshold = 90%
    Adaptive Top-K selection stops once 90% of total contribution is covered; chosen by hand.
assumptions (5)
  • domain assumption The 3DGS reconstruction of each UAV scene is accurate enough that projected Gaussian footprints align with image regions.
    Invoked in Eq. 6 and the region pooling step: 3D points are projected into 2D SAM regions, requiring precise geometry and camera poses.
  • domain assumption SAM masks provide semantically meaningful regions for the target objects in UAV views.
    Regions R_{v,r} are used as alignment anchors for both 2D feature extraction and 3D pooling (Section 'Region-based 2D-3D Feature Fusion').
  • domain assumption RemoteCLIP and RS5M features are complementary and together cover the semantic categories used in evaluation.
    Fusion in Eq. 5 assumes the two encoders additively improve recognition; no evidence is given that their combination spans the annotation vocabulary.
  • domain assumption A MinkUNet pretrained with SegContrast on point clouds, when applied to downsampled Gaussian centers and opacities, yields useful structural features.
    The 3D structural branch uses psi_3D on sparse voxel input (x_k = [p_k, o_id]) and assumes the pretrained weights transfer to this derived point cloud.
  • standard math Cosine similarity between rendered semantic features and text embeddings is an appropriate measure for segmentation and localization.
    Eq. 4 defines the heatmap H_q(v) via cosine similarity; this is the standard convention in the cited CLIP-based methods.

how reviews work

0 comments
Cite this review

Pith. "Pith review of OutLangSplat: 3D Language Gaussian Splatting for UAV Outdoor Scenes." pith.science (2026). https://pith.science/paper/VHON34YC

@misc{pith2026260804560,
  author       = {Pith},
  title        = {Pith review of: OutLangSplat: 3D Language Gaussian Splatting for UAV Outdoor Scenes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VHON34YC}},
  note         = {Machine review of arXiv:2608.04560}
}
read the original abstract

3D Language Gaussian Splatting embeds open-vocabulary language features into 3D Gaussian Splatting, providing an efficient explicit representation for text-driven 3D scene understanding. However, existing methods are limited to indoor or small-scale scenes, and tend to fail in Unmanned Aerial Vehicle (UAV) outdoor scenes, where severe occlusions and long distance viewpoints often lead to incorrect semantic activations and missing target responses. In this paper, we present OutLangSplat which adapts language Gaussian representations to UAV outdoor scenes by improving feature representation and aggregation reliability. For the feature representation, a 2D-3D dual-branch representation with region-based alignment and fusion is designed to improve spatial consistency, reducing incomplete target responses and background misactivations. For the feature aggregation, we introduce a training-free contribution and consistency-aware Gaussian feature aggregation strategy that leverages pixel contribution reliability and cross-view semantic consistency to suppress unreliable responses from noisy viewpoints. A new dataset is provided by manually annotating various objects on four real-world public UAV outdoor scene datasets. To the best of our knowledge, it is the first accessible dataset of open-vocabulary 3D scene understanding for UAV outdoor scenes. Quantitative evaluations and ablation studies demonstrate that OutLangSplat outperforms SOTA methods on both open-vocabulary semantic segmentation and instance localization tasks. The datasets and codes will be open-sourced.

Figures

Figures reproduced from arXiv: 2608.04560 by the authors.

Figure 1
Figure 1. Overview of our OutLangSplat framework. Region-level 2D semantic and 3D structural features are fused, and then [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Pixel reliability measured by the inverse Simpson [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Qualitative comparison of open-vocabulary semantic segmentation. Our method produces more complete regions, [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Qualitative comparison on open-vocabulary [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Ablation study on feature fusion for the queries [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Aggregation ablation for the query “blue roof build [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 32 canonical work pages

  1. [1]

    ACM Transactions on Graphics (TOG) , year=

    3D Gaussian Splatting for Real-Time Radiance Field Rendering , author=. ACM Transactions on Graphics (TOG) , year=

    work page

  2. [2]

    2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks , author=. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    work page 2019

  3. [3]

    2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation , author=. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    work page 2017

  4. [4]

    2019 IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    KPConv: Flexible and Deformable Convolution for Point Clouds , author=. 2019 IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    work page 2019

  5. [5]

    2024 , url=

    Occam's LGS: An Efficient Approach for Language Gaussian Splatting , author=. 2024 , url=

    work page 2024

  6. [6]

    2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    LangSplat: 3D Language Gaussian Splatting , author=. 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    work page 2024

  7. [7]

    ArXiv , year=

    OpenGaussian: Towards Point-Level 3D Gaussian-based Open Vocabulary Understanding , author=. ArXiv , year=

    work page

  8. [8]

    Splat: Directly Referring 3D Gaussian Splatting via Direct Language Embedding Registration , author=

    Dr. Splat: Directly Referring 3D Gaussian Splatting via Direct Language Embedding Registration , author=. 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    work page 2025

Show all 35 references
  1. [9]

    2023 IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    Segment Anything , author=. 2023 IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    2023

  2. [10]

    ArXiv , year=

    Learning Transferable Visual Models From Natural Language Supervision , author=. ArXiv , year=

  3. [11]

    IEEE Transactions on Geoscience and Remote Sensing , year=

    RS5M and GeoRSCLIP: A Large-Scale Vision- Language Dataset and a Large Vision-Language Model for Remote Sensing , author=. IEEE Transactions on Geoscience and Remote Sensing , year=

  4. [12]

    IEEE Transactions on Geoscience and Remote Sensing , year=

    RemoteCLIP: A Vision Language Foundation Model for Remote Sensing , author=. IEEE Transactions on Geoscience and Remote Sensing , year=

  5. [13]

    IEEE Robotics and Automation Letters , year=

    SegContrast: 3D Point Cloud Feature Representation Learning Through Self-Supervised Segment Discrimination , author=. IEEE Robotics and Automation Letters , year=

  6. [14]

    IEEE Transactions on Geoscience and Remote Sensing , year=

    3D Instance Segmentation of MVS Buildings , author=. IEEE Transactions on Geoscience and Remote Sensing , year=

  7. [15]

    European Conference on Computer Vision , year=

    Capturing, Reconstructing, and Simulating: The UrbanScene3D Dataset , author=. European Conference on Computer Vision , year=

  8. [16]

    2025 IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    LUDVIG: Learning-Free Uplifting of 2D Visual Features to Gaussian Splatting Scenes , author=. 2025 IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    2025

  9. [17]

    2026 International Conference on 3D Vision (3DV) , year=

    Visibility-Aware Language Aggregation for Open-Vocabulary Segmentation in 3D Gaussian Splatting , author=. 2026 International Conference on 3D Vision (3DV) , year=

    2026

  10. [18]

    2023 IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    LERF: Language Embedded Radiance Fields , author=. 2023 IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    2023

  11. [19]

    ArXiv , year=

    OpenNeRF: Open Set 3D Neural Scene Segmentation with Pixel-Wise Features and Rendered Novel Views , author=. ArXiv , year=

  12. [20]

    Communications of the ACM , volume=

    Nerf: Representing scenes as neural radiance fields for view synthesis , author=. Communications of the ACM , volume=. 2021 , publisher=

    2021

  13. [21]

    2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    Feature 3DGS: Supercharging 3D Gaussian Splatting to Enable Distilled Feature Fields , author=. 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    2024

  14. [22]

    2025 IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    Votesplat: Hough Voting Gaussian Splatting for 3D Scene Understanding , author=. 2025 IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    2025

  15. [23]

    2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    InstanceGaussian: Appearance-Semantic Joint Gaussian Representation for 3D Instance-Level Perception , author=. 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    2025

  16. [24]

    Proceedings

    Neural Feature Fusion Fields: 3D Distillation of Self-Supervised 2D Image Representations , author=. Proceedings. International Conference on 3D Vision , year=

  17. [25]

    ArXiv , year=

    Decomposing NeRF for Editing via Feature Field Distillation , author=. ArXiv , year=

  18. [26]

    Proceedings of the SIGGRAPH Asia 2025 Conference Papers , year=

    Lang3D-XL: Language Embedded 3D Gaussians for Large-scale Scenes , author=. Proceedings of the SIGGRAPH Asia 2025 Conference Papers , year=

    2025

  19. [27]

    Proceedings of the IEEE/CVF international conference on computer vision , pages=

    Emerging properties in self-supervised vision transformers , author=. Proceedings of the IEEE/CVF international conference on computer vision , pages=

  20. [28]

    European Conference on Computer Vision , year=

    Gaussian Grouping: Segment and Edit Anything in 3D Scenes , author=. European Conference on Computer Vision , year=

  21. [29]

    ArXiv , year=

    FAST-Splat: Fast, Ambiguity-Free Semantics Transfer in Gaussian Splatting , author=. ArXiv , year=

  22. [30]

    2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    Language Embedded 3D Gaussians for Open-Vocabulary Scene Understanding , author=. 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    2024

  23. [31]

    Proceedings of the 33rd ACM International Conference on Multimedia , year=

    SLGaussian: Fast Language Gaussian Splatting in Sparse Views , author=. Proceedings of the 33rd ACM International Conference on Multimedia , year=

  24. [32]

    2026 International Conference on 3D Vision (3DV) , year=

    SuperGSeg: Open-Vocabulary 3D Segmentation with Structured Super-Gaussians , author=. 2026 International Conference on 3D Vision (3DV) , year=

    2026

  25. [33]

    2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) , year=

    OpenSplat3D: Open-Vocabulary 3D Instance Segmentation using Gaussian Splatting , author=. 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) , year=

    2025

  26. [34]

    Proceedings of the 32nd ACM International Conference on Multimedia , year=

    GOI: Find 3D Gaussians of Interest with an Optimizable Open-vocabulary Semantic-space Hyperplane , author=. Proceedings of the 32nd ACM International Conference on Multimedia , year=

  27. [35]

    2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    PanoGS: Gaussian-based Panoptic Segmentation for 3D Open Vocabulary Scene Understanding , author=. 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    2025

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.