Pith. sign in

REVIEW 4 major objections 5 minor 4 cited by

InstanceGaussian: Appearance-Semantic Joint Gaussian Representation for 3D Instance-Level Perception

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Pith's one-line read InstanceGaussian jointly learns appearance and instance features in 3D Gaussian Splatting and aggregates instances bottom-up, reaching state-of-the-art category-agnostic segmentation on ScanNet.

desk verdict Plausible incremental method with a serious evaluation-pipeline ambiguity (the MLP smoothing step in Appendix A.3) that must be resolved before the SOTA claim is trusted. read the letter →

arxiv 2411.19235 v2 pith:LT2BV6GD submitted 2024-11-28 cs.CV

classification cs.CV
keywords 3DGaussianSplattinginstancesegmentationopen-vocabularyunderstandingSemantic-Scaffold-GSbottom-upaggregationappearance-semanticjointlearningcategory-agnostic
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

InstanceGaussian tries to make 3D Gaussian Splatting scenes usable for perception, not just rendering, by learning appearance and instance semantics in one representation. It argues that many Gaussians are needed for texture but only one semantic label is needed per region, so it gives each scaffold anchor a single instance feature shared by its five child Gaussians. It couples this with a progressive training schedule that starts with appearance, then adds semantics, then trains both together, and with a bottom-up aggregation step that over-segments the scene and merges sub-objects by feature similarity and spatial adjacency. The payoff, if the method is right, is category-agnostic instance segmentation and open-vocabulary text queries directly on 3D points, with reported gains over prior Gaussian-based approaches on ScanNet.

What carries the argument

The load-bearing object is the anchor-level shared instance feature. In Scaffold-GS, anchors generate child Gaussians; InstanceGaussian gives each anchor one instance feature f that all five of its children inherit, so semantics is learned once per region while appearance remains per-child. The second mechanism is the aggregation graph: nodes are the over-segmented sub-objects from farthest-point-sampling plus k-means, and an edge exists when the L2 distance between sub-object features is below a threshold and the sub-objects share a voxel in a voxelized space; connected components then define the final instances. The third mechanism is the progressive schedule, which starts with appearance-only training, then separates appearance and semantic optimization, then jointly trains both, preventing semantic features from being fitted to a frozen appearance field.

What would settle it

Run the same pipeline with per-child instance features instead of a shared anchor feature and compare boundary-region segmentation quality on ScanNet; if per-child features improve boundary IoU, the shared-feature assumption is the limiting factor. Alternatively, take a scene where all SAM masks are correct and check whether the large-object aggregation failure persists, which would rule mask quality out as the cause.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that separating the granularity of appearance from the granularity of semantics inside a Gaussian scene removes a bottleneck in Gaussian-based perception. In the proposed Semantic-Scaffold-GS representation, each parent anchor holds an appearance embedding and one instance feature, and the five child Gaussians decoded from that anchor share the feature while keeping their own color, position, scale and rotation. Appearance and semantics are then trained in three phases so that neither is frozen while the other learns, using an intra-mask smoothness loss and a contrastive loss truncated at a distance threshold to avoid destabilizing joint training. Instance segmentation is produced bottom-up: farthest point sampling seeds about a thousand sub-object clusters, k-means groups the Gaussians in position-plus-feature space, and a graph whose edges combine feature distance with voxel adjacency merges sub-objects into complete objects with no preset category or object count. On the ten ScanNet scenes used by the baseline, the paper reports 50.27 instance mIoU against 27.32 for OpenGaussian, and 47.94 versus 38.29 for ten-class open-vocabulary semantic segmentation.

Load-bearing premise

The argument assumes each scaffold anchor lies within a single object, so the shared instance feature inherited by its five child Gaussians cannot blur object boundaries; an anchor straddling two objects sacrifices boundary precision.

Editorial extensions

If this is right

  • Category-agnostic instance segmentation on Gaussians no longer requires knowing how many objects are in the scene; the graph aggregation step determines the count adaptively.
  • A single trained scene supports rendering, instance segmentation, and open-vocabulary text retrieval from the same learned features, as the paper's LeRF object-selection experiments demonstrate.
  • The method inherits a dependency on SAM mask quality, because the instance-feature losses are supervised by those masks; the paper reports that frequent SAM failures degrade segmentation accuracy.
  • Training time increases relative to the baselines (about 1847 seconds per ScanNet scene in the paper's measurements) because appearance attributes are decoded through MLPs.

Reading between the lines

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

  • The paper leaves the shared-per-anchor feature untested: it does not ablate the child count, so a per-child feature variant could preserve or improve boundary precision when anchors cross object edges.
  • The bottom-up aggregation is modular and could be applied to other Gaussian feature fields, isolating how much of the gain comes from aggregation versus the joint representation.
  • The reported large-object failure points to a concrete extension: weight the smoothness loss by cross-view observation completeness so rarely seen parts are not pushed apart by the contrastive term.
  • A mask-free self-supervised consistency signal could replace SAM supervision in scenes where masks are unreliable, testing how much the method depends on mask quality.
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 / 5 minor

Summary. The paper proposes InstanceGaussian, a 3D Gaussian Splatting-based method for joint appearance-semantic representation and bottom-up category-agnostic instance segmentation. It introduces a Semantic-Scaffold-GS representation in which anchor-level instance features are shared among child Gaussians, a progressive appearance-semantic joint training strategy, and a graph-connectivity aggregation algorithm built on farthest-point-sampling over-segmentation. Experiments on ScanNet and LeRF report large improvements over Gaussian-based baselines, including OpenGaussian and GaussianGrouping, for instance segmentation, open-vocabulary point cloud understanding, and object selection/rendering.

Significance. If the reported results are reproducible and correctly attributed, the paper makes a useful advance: the anchor-level shared instance feature and the bottom-up aggregation are clean, simple ideas that improve both accuracy and boundary quality over prior Gaussian-based methods. The paper includes helpful ablations (Tables 4, 5, and supplementary C), a hyperparameter robustness study (Table 10), a failure-case analysis, and a comparison with additional baselines in the supplementary material. These are strengths. However, the strength of the central 'state-of-the-art' claim is currently weakened by an ambiguously described post-processing step in the ScanNet evaluation pipeline and by the absence of variance estimates in the main tables.

major comments (4)
  1. [Appendix A.3] The MLP smoothing step described in Appendix A.3 ('we identified well-optimized point clouds based on their contributions during rendering and utilized them to train a lightweight MLP... the trained MLP is employed to infer segmentation results for the point clouds') is potentially load-bearing for Tables 1, 2, 4, and 5, since those tables report ScanNet point-level mIoU/mAcc. The manuscript does not state whether the reported numbers come from the raw Sec. 3.3 output or from the MLP-inferred labels, and there is no ablation of this step. The authors must explicitly clarify the evaluation protocol and, if the MLP output is used, provide an ablation that isolates its contribution. Without this, the reported gains over OpenGaussian cannot be attributed to the proposed representation and aggregation method.
  2. [Tables 1-3, C.3] All main results are single runs without variance, yet the algorithm depends on randomized FPS initialization and k-means. The paper's own supplementary Table 11 reports that five FPS random initializations yield semantic segmentation mIoU 39.44±1.38 and instance segmentation mIoU 48.57±0.98, indicating that the point estimates in the main tables (for example 50.27 in Table 1) have nontrivial uncertainty. The authors should report mean±std over at least a few runs for the main tables, or explicitly justify why single runs are sufficient.
  3. [Sec. 4.2, Table 2; Sec. 4.3, Table 3] The comparison set is too narrow for the claim of state-of-the-art performance. Related works discussed in Sec. 2, including SAGA, GAGA, ClickGaussian, and Feature3DGS, are not evaluated in the main experiments; only the supplementary material (Table 7) adds GAGA and SAGA, without explaining why these are absent from the main comparison or how their metrics were obtained. The authors should either include all feasible Gaussian-based baselines in the main tables or justify their exclusion in terms of input modalities, task definitions, or code availability.
  4. [Sec. 3.2(1)] The central architectural assumption that five child Gaussians share a single anchor-level instance feature is never ablated. The paper does not report results for different child counts (e.g., 1, 3, 10) or for a variant where each child has an independent instance feature. Since this assumption directly affects boundary precision at object edges, the authors should include an ablation over child count or an independence variant to support the claim that the shared anchor feature is beneficial.
minor comments (5)
  1. [Sec. 4.1 and Appendix A.3] The paper says in Sec. 4.1 'the 10 scenes selected by OpenGaussian' but in A.3 says 'We randomly selected 10 scenes from ScanNet for evaluation.' These statements are inconsistent and should be reconciled.
  2. [Table 1 caption] The phrase 'Accurate is measured by mAcc@0.25' is unclear. Please define what mAcc@0.25 means (e.g., whether instances are matched at a 25% IoU threshold) and how it is computed.
  3. [Eq. (4) and Algorithm 1] The output ordering of Eq. (4) is (f̄, μ̄, Ī) while Algorithm 1 line 4 writes (Ī, f̄, μ̄). Please make the notation consistent.
  4. [Fig. 2] The stage labels '0-10k', '10-20k', '20-30k' should specify the unit (training iterations) and whether these are global steps or per-stage steps.
  5. [Table 6 caption] Typo: 'Comparation' should be 'Comparison'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central SOTA claims are benchmarked against external ScanNet/LeRF ground truth, and the reused OpenGaussian losses are explicitly stated rather than circularly cited.

full rationale

The paper's central claims are validated on external ScanNet and LeRF annotations, so the reported mIoU/mAcc numbers do not reduce to a fitted constant or to the method's own definitions. Reusing the authors' prior OpenGaussian smooth/contrastive losses and its CLIP-association pipeline is a self-extension, not a circular load: the losses are written out in Eqs. (1)-(3), and the final metrics depend on ground-truth labels rather than on accepting OpenGaussian's results on faith. The Proposed Semantic-Scaffold-GS representation, progressive training schedule, and bottom-up FPS/k-means/graph aggregation are all described operationally, with ablations over the main design choices (Tables 4-5). One reporting gap should be flagged separately: Appendix A.3 describes a lightweight MLP, trained on 'well-optimized point clouds', that takes position and color and 'predicts segmentation labels as outputs' for the point clouds. If this MLP is part of the evaluated pipeline, the reported gains would need an ablation to attribute them to the proposed representation; however, this is an evaluation-attribution and reproducibility concern, not a circular reduction by construction, because the MLP is not used to define the instance features or the aggregation objective and no equation equates the final metric to the training target.

Assumptions & free parameters 6 free parameters · 4 assumptions · 1 invented entities

The method's claims are empirical. The main load-bearing choices are architectural (shared anchor semantic features), supervision (SAM masks), and evaluation (OpenGaussian's CLIP association). Several thresholds and cluster counts were hand-selected, and no code is released to verify the exact pipeline.

free parameters (6)
  • Contrastive loss truncation threshold tau = 0.4
    Hand-set in Eq. 3 to stabilize joint training; reported with no sensitivity analysis.
  • Connectivity threshold gamma = 0.1
    Hand-set in Algorithm 1; ablation in Table 10 covers 0.06 to 0.18.
  • Voxel resolution r = 0.05 to 0.5 (range)
    Selected by hand for voxelization; ablation covers a range but no default is stated.
  • Number of FPS over-segment seeds s = 1000 default
    Sec. 3.3 default value; ablation from 200 to 1000 changes instance mIoU by 49.60 +/- 1.31.
  • Child Gaussians per anchor = 5
    Architectural choice in Semantic-Scaffold-GS; no ablation of child count is provided.
  • Instance feature dimension = 6
    Inherited from OpenGaussian; no ablation or justification for this dimension.
assumptions (4)
  • domain assumption The five child Gaussians derived from one anchor are semantically coherent, so sharing one instance feature per anchor does not blur object boundaries.
    Section 3.2(1) builds Semantic-Scaffold-GS on this premise; if an anchor straddles multiple objects, the shared feature cannot represent the boundary.
  • domain assumption SAM masks are reliable supervision for instance features.
    The losses Ls and Lc in Eqs. 1 to 3 use SAM masks; the paper's own failure analysis states that frequent mask failures degrade segmentation accuracy.
  • domain assumption OpenGaussian's 2D mask to 3D instance association and CLIP features are adequate for open-vocabulary evaluation.
    Section 4.2 adopts this external pipeline; errors in that association would limit open-vocabulary results.
  • standard math k-means and connected-component algorithms on the feature plus position space produce meaningful over-segmentation and aggregation.
    Algorithm 1 relies on standard clustering and graph traversal; no formal guarantee links learned feature geometry to object instances.
invented entities (1)
  • Anchor-level shared instance feature f
    purpose: Give all child Gaussians of one anchor the same semantic representation, balancing appearance and semantics.
    New latent variable introduced by the paper; no falsifiable handle outside the downstream segmentation benchmarks.

how reviews work

0 comments
Cite this review

Pith. "Pith review of InstanceGaussian: Appearance-Semantic Joint Gaussian Representation for 3D Instance-Level Perception." pith.science (2026). https://pith.science/paper/LT2BV6GD

@misc{pith2026241119235,
  author       = {Pith},
  title        = {Pith review of: InstanceGaussian: Appearance-Semantic Joint Gaussian Representation for 3D Instance-Level Perception},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LT2BV6GD}},
  note         = {Machine review of arXiv:2411.19235}
}
read the original abstract

3D scene understanding has become an essential area of research with applications in autonomous driving, robotics, and augmented reality. Recently, 3D Gaussian Splatting (3DGS) has emerged as a powerful approach, combining explicit modeling with neural adaptability to provide efficient and detailed scene representations. However, three major challenges remain in leveraging 3DGS for scene understanding: 1) an imbalance between appearance and semantics, where dense Gaussian usage for fine-grained texture modeling does not align with the minimal requirements for semantic attributes; 2) inconsistencies between appearance and semantics, as purely appearance-based Gaussians often misrepresent object boundaries; and 3) reliance on top-down instance segmentation methods, which struggle with uneven category distributions, leading to over- or under-segmentation. In this work, we propose InstanceGaussian, a method that jointly learns appearance and semantic features while adaptively aggregating instances. Our contributions include: i) a novel Semantic-Scaffold-GS representation balancing appearance and semantics to improve feature representations and boundary delineation; ii) a progressive appearance-semantic joint training strategy to enhance stability and segmentation accuracy; and iii) a bottom-up, category-agnostic instance aggregation approach that addresses segmentation challenges through farthest point sampling and connected component analysis. Our approach achieves state-of-the-art performance in category-agnostic, open-vocabulary 3D point-level segmentation, highlighting the effectiveness of the proposed representation and training strategies. Project page: https://lhj-git.github.io/InstanceGaussian/

Figures

Figures reproduced from arXiv: 2411.19235 by the authors.

Figure 1
Figure 1. Top row: Appearance-semantic joint Gaussian representation avoids the imbalance and inconsistency in appearance-semantic [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Progressive appearance-semantic joint training. (a) Train [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Visualization comparison of category-agnostic 3D instance segmentation result. InstanceGaussian outperforms OpenGaussian [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: Open-vocabulary query point cloud Understanding on ScanNet dataset. InstanceGaussian shows advanced text query capabilities. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Open-vocabulary 3D object selection and rendering on the LeRF dataset. InstanceGaussian outperforms OpenGaussian in [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Visualization results of the appearance color and instance features of the Gaussians ellipsoid in the 3D viewer demonstrate that [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Top row: Reference mesh of scenes. Bottom row: The visualization result of category-agnostic 3D instance segmentation in [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]
Figure 8
Figure 8. Figure 8: The visualization result of open-vocabulary query point cloud understanding in ScanNet [ [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]
Figure 9
Figure 9. Figure 9: The 2D visualization result of 3D instance segmentation in LeRF dataset. [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 10
Figure 10. Figure 10: Top: Reference image of scenes. Middle: Constructed 3D Gaussians/points. Bottom: The visualization result of category [PITH_FULL_IMAGE:figures/full_fig_p013_10.png]
Figure 11
Figure 11. Figure 11: The feature map of InstanceGaussian. Left: Reference images of the scene. Middle: The visualization result of feature maps of [PITH_FULL_IMAGE:figures/full_fig_p013_11.png]
Figure 12
Figure 12. Figure 12: The failure case of ramen in LeRF dataset. Frequent mask failures will undermine the ability to distinguish different food in the [PITH_FULL_IMAGE:figures/full_fig_p014_12.png]
Figure 13
Figure 13. Figure 13: The failure case of aggregate the large meeting table in scene0140 [PITH_FULL_IMAGE:figures/full_fig_p014_13.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 4 Pith papers

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

  1. IGFuse: Interactive 3D Gaussian Scene Reconstruction via Multi-Scans Fusion

    cs.CV 2025-08 conditional novelty 6.0 of 10

    IGFuse jointly optimizes segmentation-aware Gaussian fields from multiple scans of rearranged scenes, producing complete, manipulable 3D reconstructions without inpainting.

  2. VoteSplat: Hough Voting Gaussian Splatting for 3D Scene Understanding

    cs.GR 2025-06 conditional novelty 6.0 of 10

    VoteSplat embeds per-Gaussian 3D offset vectors, supervises them with SAM mask centers, and clusters the resulting 3D votes to segment and localize objects in Gaussian Splatting scenes.

  3. DSG-World: Learning a 3D Gaussian World Model from Dual State Videos

    cs.CV 2025-06 conditional novelty 6.0 of 10

    DSG-World builds two segmented 3D Gaussian fields from two scene states and trains them with mutual consistency, enabling novel-state simulation without inpainting or dense capture.

  4. OGGSplat: Open Gaussian Growing for Generalizable Reconstruction with Expanded Field-of-View

    cs.CV 2025-06 conditional novelty 6.0 of 10

    A method that grows open-vocabulary 3D Gaussians beyond the input view cone by bidirectionally consistent RGB and semantic diffusion inpainting.

Reference graph

Works this paper leans on

45 extracted references · 41 canonical work pages · cited by 4 Pith papers

  1. [1]

    Neural point-based graph- ics

    Kara-Ali Aliev, Artem Sevastopolsky, Maria Kolos, Dmitry Ulyanov, and Victor Lempitsky. Neural point-based graph- ics. In European Conference on Computer Vision (ECCV) ,

  2. [2]

    Mip-nerf: A multiscale representation for anti-aliasing neu- ral radiance fields

    Jonathan T Barron, Ben Mildenhall, Matthew Tancik, Peter Hedman, Ricardo Martin-Brualla, and Pratul P Srinivasan. Mip-nerf: A multiscale representation for anti-aliasing neu- ral radiance fields. In Proceedings of the IEEE/CVF Interna- tional Conference on Computer Vision (ICCV), 2021. 2

  3. [3]

    Mip-nerf 360: Unbounded anti-aliased neural radiance fields

    Jonathan T Barron, Ben Mildenhall, Dor Verbin, Pratul P Srinivasan, and Peter Hedman. Mip-nerf 360: Unbounded anti-aliased neural radiance fields. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022

  4. [4]

    Zip-nerf: Anti-aliased grid-based neural radiance fields

    Jonathan T Barron, Ben Mildenhall, Dor Verbin, Pratul P Srinivasan, and Peter Hedman. Zip-nerf: Anti-aliased grid-based neural radiance fields. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2023. 2

  5. [5]

    Density-based clustering based on hierarchical density esti- mates

    Ricardo JGB Campello, Davoud Moulavi, and J ¨org Sander. Density-based clustering based on hierarchical density esti- mates. In Pacific-Asia conference on knowledge discovery and data mining, pages 160–172, 2013. 10

  6. [6]

    Emerg- ing properties in self-supervised vision transformers

    Mathilde Caron, Hugo Touvron, Ishan Misra, Herv ´e J´egou, Julien Mairal, Piotr Bojanowski, and Armand Joulin. Emerg- ing properties in self-supervised vision transformers. In Pro- ceedings of the IEEE/CVF international conference on com- puter vision (ICCV), 2021. 2

  7. [7]

    Segment any 3d gaussians

    Jiazhong Cen, Jiemin Fang, Chen Yang, Lingxi Xie, Xi- aopeng Zhang, Wei Shen, and Qi Tian. Segment any 3d gaussians. arXiv preprint arXiv:2312.00860 , 2023. 2, 3, 9

  8. [8]

    Click-gaussian: Interactive segmenta- tion to any 3d gaussians

    Seokhun Choi, Hyeonseop Song, Jaechul Kim, Taehyeong Kim, and Hoseok Do. Click-gaussian: Interactive segmenta- tion to any 3d gaussians. In European Conference on Com- puter Vision (ECCV), 2024. 2, 3

Show all 45 references
  1. [9]

    Scannet: Richly-annotated 3d reconstructions of indoor scenes

    Angela Dai, Angel X Chang, Manolis Savva, Maciej Hal- ber, Thomas Funkhouser, and Matthias Nießner. Scannet: Richly-annotated 3d reconstructions of indoor scenes. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017. 5, 9, 10, 11

  2. [10]

    Graspnet-1billion: A large-scale benchmark for general ob- ject grasping

    Hao-Shu Fang, Chenxi Wang, Minghao Gou, and Cewu Lu. Graspnet-1billion: A large-scale benchmark for general ob- ject grasping. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020. 10

  3. [11]

    Efficient graph-based image segmentation

    Pedro F Felzenszwalb and Daniel P Huttenlocher. Efficient graph-based image segmentation. International journal of computer vision, 2004. 3

  4. [12]

    Plenoxels: Radiance fields without neural networks

    Sara Fridovich-Keil, Alex Yu, Matthew Tancik, Qinhong Chen, Benjamin Recht, and Angjoo Kanazawa. Plenoxels: Radiance fields without neural networks. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022. 2

  5. [13]

    Hicom: Hierarchical coherent motion for dy- namic streamable scenes with 3d gaussian splatting

    Qiankun Gao, Jiarui Meng, Chengxiang Wen, Jie Chen, and Jian Zhang. Hicom: Hierarchical coherent motion for dy- namic streamable scenes with 3d gaussian splatting. In Ad- vances in Neural Information Processing Systems (NeurIPS),

  6. [14]

    2d gaussian splatting for geometrically ac- curate radiance fields

    Binbin Huang, Zehao Yu, Anpei Chen, Andreas Geiger, and Shenghua Gao. 2d gaussian splatting for geometrically ac- curate radiance fields. In ACM SIGGRAPH 2024 Conference Papers, 2024. 2

  7. [15]

    Clip2point: Transfer clip to point cloud classifica- tion with image-depth pre-training

    Tianyu Huang, Bowen Dong, Yunhan Yang, Xiaoshui Huang, Rynson WH Lau, Wanli Ouyang, and Wangmeng Zuo. Clip2point: Transfer clip to point cloud classifica- tion with image-depth pre-training. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2023. 2

  8. [16]

    Fastlgs: Speeding up lan- guage embedded gaussians with feature grid mapping

    Yuzhou Ji, He Zhu, Junshu Tang, Wuyi Liu, Zhizhong Zhang, Xin Tan, and Yuan Xie. Fastlgs: Speeding up lan- guage embedded gaussians with feature grid mapping. In Proceedings of the AAAI Conference on Artificial Intelli- gence, 2025. 1

  9. [17]

    Alignerf: High-fidelity neural radiance fields via alignment- aware training

    Yifan Jiang, Peter Hedman, Ben Mildenhall, Dejia Xu, Jonathan T Barron, Zhangyang Wang, and Tianfan Xue. Alignerf: High-fidelity neural radiance fields via alignment- aware training. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023. 2

  10. [18]

    3d gaussian splatting for real-time radiance field rendering

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimk ¨uhler, and George Drettakis. 3d gaussian splatting for real-time radiance field rendering. ACM Transactions on Graphics (TOG), 2023. 1, 2, 3, 9

  11. [19]

    Lerf: Language embedded radiance fields

    Justin Kerr, Chung Min Kim, Ken Goldberg, Angjoo Kanazawa, and Matthew Tancik. Lerf: Language embedded radiance fields. In Proceedings of the IEEE/CVF Interna- tional Conference on Computer Vision (ICCV) , 2023. 2, 7, 9, 10

  12. [20]

    Segment any- thing

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer White- head, Alexander C Berg, Wan-Yen Lo, et al. Segment any- thing. In Proceedings of the IEEE/CVF International Con- ference on Computer Vision (ICCV), 2023. 2, 9

  13. [21]

    Weakly supervised 3d open- vocabulary segmentation

    Kunhao Liu, Fangneng Zhan, Jiahui Zhang, Muyu Xu, Yingchen Yu, Abdulmotaleb El Saddik, Christian Theobalt, Eric Xing, and Shijian Lu. Weakly supervised 3d open- vocabulary segmentation. In NeurIPS, 2023. 2

  14. [22]

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

    Tao Lu, Mulin Yu, Linning Xu, Yuanbo Xiangli, Limin Wang, Dahua Lin, and Bo Dai. Scaffold-gs: Structured 3d gaussians for view-adaptive rendering. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024. 2, 3

  15. [23]

    Gaga: Group any gaussians via 3d-aware memory bank, 2024

    Weijie Lyu, Xueting Li, Abhijit Kundu, Yi-Hsuan Tsai, and Ming-Hsuan Yang. Gaga: Group any gaussians via 3d-aware memory bank, 2024. 9

  16. [24]

    Mirror-3dgs: Incorporat- ing mirror reflections into 3d gaussian splatting

    Jiarui Meng, Haijie Li, Yanmin Wu, Qiankun Gao, Shuzhou Yang, Jian Zhang, and Siwei Ma. Mirror-3dgs: Incorporat- ing mirror reflections into 3d gaussian splatting. In 2024 IEEE International Conference on Visual Communications and Image Processing (VCIP), pages 1–5. IEEE, 2024. 2

  17. [25]

    Nerf: Representing scenes as neural radiance fields for view syn- thesis

    Ben Mildenhall, Pratul P Srinivasan, Matthew Tancik, Jonathan T Barron, Ravi Ramamoorthi, and Ren Ng. Nerf: Representing scenes as neural radiance fields for view syn- thesis. Communications of the ACM, 2021. 2

  18. [26]

    Instant neural graphics primitives with a mul- tiresolution hash encoding

    Thomas M ¨uller, Alex Evans, Christoph Schied, and Alexan- der Keller. Instant neural graphics primitives with a mul- tiresolution hash encoding. ACM transactions on graphics (TOG), 2022. 2

  19. [27]

    Open3dis: Open-vocabulary 3d instance segmentation with 2d mask guidance

    Phuc Nguyen, Tuan Duc Ngo, Evangelos Kalogerakis, Chuang Gan, Anh Tran, Cuong Pham, and Khoi Nguyen. Open3dis: Open-vocabulary 3d instance segmentation with 2d mask guidance. In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (CVPR) ,

  20. [28]

    Langsplat: 3d language gaussian splatting

    Minghan Qin, Wanhua Li, Jiawei Zhou, Haoqian Wang, and Hanspeter Pfister. Langsplat: 3d language gaussian splatting. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 20051–20060, 2024. 1, 2, 4, 6, 7

  21. [29]

    Learning transferable visual models from natural language supervi- sion

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervi- sion. In International conference on machine learning, 2021. 2

  22. [30]

    Merf: Memory-efficient radiance fields for real- time view synthesis in unbounded scenes.ACM Transactions on Graphics (TOG), 2023

    Christian Reiser, Rick Szeliski, Dor Verbin, Pratul Srini- vasan, Ben Mildenhall, Andreas Geiger, Jon Barron, and Pe- ter Hedman. Merf: Memory-efficient radiance fields for real- time view synthesis in unbounded scenes.ACM Transactions on Graphics (TOG), 2023. 2

  23. [31]

    Adop: Approximate differentiable one-pixel point rendering

    Darius R ¨uckert, Linus Franke, and Marc Stamminger. Adop: Approximate differentiable one-pixel point rendering. ACM TOG, 2022. 2

  24. [32]

    Language embedded 3d gaussians for open- vocabulary scene understanding

    Jin-Chuan Shi, Miao Wang, Hao-Bin Duan, and Shao- Hua Guan. Language embedded 3d gaussians for open- vocabulary scene understanding. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5333–5343, 2024. 2, 6, 7

  25. [33]

    Direct voxel grid optimization: Super-fast convergence for radiance fields reconstruction

    Cheng Sun, Min Sun, and Hwann-Tzong Chen. Direct voxel grid optimization: Super-fast convergence for radiance fields reconstruction. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022. 2

  26. [34]

    Differentiable surface splatting for point-based geometry processing

    Yifan Wang, Felice Serena, Shihao Wu, Cengiz ¨Oztireli, and Olga Sorkine-Hornung. Differentiable surface splatting for point-based geometry processing. ACM Transactions on Graphics (TOG), 2019. 2

  27. [35]

    Eda: Explicit text-decoupling and dense alignment for 3d visual grounding

    Yanmin Wu, Xinhua Cheng, Renrui Zhang, Zesen Cheng, and Jian Zhang. Eda: Explicit text-decoupling and dense alignment for 3d visual grounding. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recogni- tion (CVPR), 2023. 2

  28. [36]

    Language-assisted 3d scene understanding

    Yanmin Wu, Qiankun Gao, Renrui Zhang, Haijie Li, and Jian Zhang. Language-assisted 3d scene understanding. IEEE Transactions on Multimedia, 2025. 2

  29. [37]

    Opengaussian: Towards point-level 3d gaussian-based open vocabulary understand- ing

    Yanmin Wu, Jiarui Meng, Haijie Li, Chenming Wu, Yahao Shi, Xinhua Cheng, Chen Zhao, Haocheng Feng, Errui Ding, Jingdong Wang, and Jian Zhang. Opengaussian: Towards point-level 3d gaussian-based open vocabulary understand- ing. In Advances in Neural Information Processing Syste...

  30. [38]

    Ulip-2: Towards scalable multimodal pre-training for 3d understanding

    Le Xue, Ning Yu, Shu Zhang, Artemis Panagopoulou, Jun- nan Li, Roberto Mart´ın-Mart´ın, Jiajun Wu, Caiming Xiong, Ran Xu, Juan Carlos Niebles, et al. Ulip-2: Towards scalable multimodal pre-training for 3d understanding. In Proceed- ings of the IEEE/CVF Conference on Computer ...

  31. [39]

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

    Chi Yan, Delin Qu, Dan Xu, Bin Zhao, Zhigang Wang, Dong Wang, and Xuelong Li. Gs-slam: Dense visual slam with 3d gaussian splatting. In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (CVPR) ,

  32. [40]

    Maskclus- tering: View consensus based mask graph clustering for open-vocabulary 3d instance segmentation

    Mi Yan, Jiazhao Zhang, Yan Zhu, and He Wang. Maskclus- tering: View consensus based mask graph clustering for open-vocabulary 3d instance segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition (CVPR), 2024. 2, 3, 5, 6

  33. [41]

    Gaussian grouping: Segment and edit anything in 3d scenes

    Mingqiao Ye, Martin Danelljan, Fisher Yu, and Lei Ke. Gaussian grouping: Segment and edit anything in 3d scenes. In European Conference on Computer Vision (ECCV), 2024. 1, 3, 5, 6

  34. [42]

    Sai3d: Segment any in- stance in 3d scenes

    Yingda Yin, Yuzheng Liu, Yang Xiao, Daniel Cohen-Or, Jingwei Huang, and Baoquan Chen. Sai3d: Segment any in- stance in 3d scenes. InProceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (CVPR) ,

  35. [43]

    Mip-splatting: Alias-free 3d gaussian splat- ting

    Zehao Yu, Anpei Chen, Binbin Huang, Torsten Sattler, and Andreas Geiger. Mip-splatting: Alias-free 3d gaussian splat- ting. In Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition (CVPR), 2024. 2

  36. [44]

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

    Shijie Zhou, Haoran Chang, Sicheng Jiang, Zhiwen Fan, Ze- hao Zhu, Dejia Xu, Pradyumna Chari, Suya You, Zhangyang Wang, and Achuta Kadambi. Feature 3dgs: Supercharging 3d gaussian splatting to enable distilled feature fields. InPro- ceedings of the IEEE/CVF Conference on Compu...

  37. [45]

    Point- clip v2: Prompting clip and gpt for powerful 3d open-world learning

    Xiangyang Zhu, Renrui Zhang, Bowei He, Ziyu Guo, Ziyao Zeng, Zipeng Qin, Shanghang Zhang, and Peng Gao. Point- clip v2: Prompting clip and gpt for powerful 3d open-world learning. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2023. 2

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.