Pith. sign in

REVIEW 4 major objections 4 minor 48 references

Planar Gaussian Splatting

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

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

Pith's one-line read A scene's 3D planes can be parsed from posed RGB video by grouping Gaussian splats, with no plane labels or depth supervision.

desk verdict Sound, well-engineered unsupervised planar reconstruction; the SOTA claim is plausible but the self-supervision loop needs an isolating ablation and the paper lacks reproducibility details. read the letter →

arxiv 2412.01931 v1 pith:S5ZSHTXV submitted 2024-12-02 cs.CV

Farhad G. Zanjani , Hong Cai , Hanno Ackermann , Leila Mirvakhabova , Fatih Porikli This is my paper · ORCID
classification cs.CV
keywords planarreconstruction3DGaussiansplattingmixturetreeplaneinstancesegmentationunsupervisedunderstandingSegmentAnythingModelsurfacenormalestimationneuralrendering
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 sets out to show that 3D plane instances in an indoor scene can be reconstructed from posed RGB images together with off-the-shelf 2D segmentation and normal maps, without any 3D plane labels or depth supervision. Instead of training a network to predict planes, Planar Gaussian Splatting (PGS) optimizes a Gaussian splatting field per scene and groups the Gaussian primitives into planes using a probabilistic tree. It does this by adding a surface normal and a learned plane descriptor to each Gaussian, lifting 2D masks from a general segmentation model into 3D through a closed-form linear regression, and merging adjacent segments whose normals and planar distances agree. The payoff, if the reported results hold, is a planar reconstruction pipeline that transfers across datasets without retraining and that the paper reports as more accurate on ScanNet and Replica than supervised baselines, while running in about 16 minutes. It also requires no depth input at all, unlike the closest optimization-based predecessor.

What carries the argument

The Gaussian Mixture Tree (GMT), a hierarchical tree over Gaussian primitives in which leaf nodes are local clusters of Gaussians, parent nodes are formed by merging nearby Gaussian distributions, and the children of the root are the recovered 3D plane instances. Two extra per-Gaussian parameters carry the argument: a surface normal, supervised by rendered normal maps against an off-the-shelf predictor, and a unit-norm plane descriptor, learned from lifted SAM masks via a per-view closed-form linear regression. Merging is gated by both the Bhattacharyya distance between Gaussian distributions and the cosine similarity of descriptors, while a recurrent mean-shift update applied to the whole field keeps descriptors for different planes separable.

What would settle it

Run PGS on a scene with a large dark wall beside a lighter wall at the same orientation and compare the recovered plane instances with the ground-truth plane boundaries; the paper's stated limitation predicts the dark wall will be under-reconstructed and split, which would directly test whether the descriptor grouping and region adjacency graph merging hold where rendered depth is weakest.

Watch

Extended reading notes

Core claim

The central claim is that plane parsing can be folded into the same optimization that reconstructs geometry, so no supervised plane predictor is needed. PGS treats each 3D Gaussian primitive as carrying a unit-norm plane descriptor and a surface normal; the descriptors are learned by rendering them into each camera view and solving a closed-form linear regression that predicts the merged 2D segment labels produced from SAM masks. A region adjacency graph then merges SAM segments that lie on the same plane using normal and planar-distance thresholds, and a tree-structured Gaussian mixture merges Gaussian clusters bottom-up, with the root's children being the final plane instances. The paper reports that this unsupervised, per-scene optimization reaches higher plane segmentation scores on ScanNet and Replica than supervised methods such as PlanarRecon, which were trained on ScanNet and degrade under domain shift, and that it outperforms the optimization-based NMF while taking 16 minutes instead of 40.

Load-bearing premise

The method assumes the depth and normal fields rendered from the Gaussian splats are accurate enough to decide which 2D segments belong to the same plane, so any region where few Gaussians are placed, such as dark areas, can degrade the plane labels that the whole grouping depends on.

Editorial extensions

If this is right

  • Plane instances come out directly from the optimized Gaussian field, so the usual post-processing chain of meshing, Marching Cubes, and Sequential RANSAC is bypassed.
  • Because the optimization is per scene, the method does not suffer the domain-shift failure mode of supervised plane predictors; the paper demonstrates this by transferring from ScanNet to Replica.
  • The reported 16-minute runtime, against 40 minutes for NMF, makes scan-to-plane reconstruction practical on a single consumer GPU.
  • Any improvement in off-the-shelf 2D segmentation or normal prediction should translate directly into better 3D plane labels within the same pipeline.
  • The method's plane count is not fixed in advance, so it can parse scenes with an arbitrary number of planar surfaces.

Reading between the lines

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

  • The closed-form lifting of 2D masks to 3D descriptors suggests PGS could be converted into an online system that updates plane labels incrementally as new frames arrive, rather than re-solving the linear regression over the full image set.
  • The same tree-merging machinery could be applied to non-planar structure by replacing the planar-distance criterion with a curvature-aware geometric agreement measure, which the paper does not explore.
  • Because the descriptors are unit-norm vectors in $\mathbb{R}^3$, the method effectively learns a mapping from Gaussian position to a low-dimensional plane embedding; one could test whether that embedding geometry corresponds directly to plane parameters such as offset and orientation.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The paper introduces Planar Gaussian Splatting (PGS), a per-scene optimization method that reconstructs 3D plane instances from posed RGB images using 3D Gaussian Splatting. The method augments each Gaussian with a surface normal and a plane descriptor, lifts 2D SAM masks into 3D via rendered depth and normals, merges them through a Region Adjacency Graph, and groups the resulting segments with a hierarchical Gaussian Mixture Tree. Experiments on 10 ScanNet scenes and Replica report state-of-the-art plane segmentation scores against NMF, PlanarRecon, and RANSAC-based baselines, along with a runtime reduction relative to NMF. An ablation study on two ScanNet scenes attributes performance gains to the SAM-based descriptors, local planar alignment, normal supervision, holistic separability, and Laplacian smoothing.

Significance. If the claims hold, PGS is a meaningful step toward RGB-only, label-free 3D planar reconstruction, combining the geometric convenience of 3D Gaussian primitives with the semantic priors of a foundation model. The per-scene optimization design sidesteps cross-dataset generalization problems that hurt supervised methods, and the reported Replica results support this advantage. The paper is clearly written and includes an ablation study that shows each proposed component contributes, plus a runtime comparison that is favorable. However, the evaluation is thin (10 ScanNet scenes, no error bars), and the core self-supervision loop—where rendered geometry creates the pseudo-labels that train the descriptors—is not controlled experimentally. These gaps currently prevent the SOTA claim from being fully established.

major comments (4)
  1. [§3.2 and Supplementary §A.2] The construction of the Gaussian Mixture Tree is underspecified: Algorithm 1 takes leaf nodes each with a normal n_i and descriptor z_i, but Supplementary A.2 states that leaf nodes are Gaussian distributions fitted to boundary points of merged SAM masks, which have no intrinsic descriptor or normal. The paper does not explain how each leaf node obtains its descriptor and normal from the underlying Gaussian primitives, nor how the plane instances represented by the child nodes of the root are mapped back to individual Gaussians to produce the final plane segmentation used in evaluation. This makes the central grouping mechanism non-reproducible and needs to be clarified.
  2. [§3.3, Eq. (7)] The pseudo-labels that supervise descriptor learning are generated by RAG partitioning that depends on the depth d(u,v) and normals rendered from the very Gaussian field being optimized. The Limitations section acknowledges that dark regions are under-reconstructed due to sparse Gaussian assignment, so these rendered quantities are biased exactly where geometry is hardest. Because no experiment isolates this dependency (e.g., replacing rendered depth/normals with ground-truth values during label generation, or measuring label quality as training progresses), the reported SOTA performance cannot yet be attributed to a robust unsupervised mechanism rather than to a self-supervision loop that reproduces its own geometry biases.
  3. [§4.1, Tables 1 and 2] The evaluation uses only 10 ScanNet scenes and a single Replica dataset, with no error bars or per-scene breakdown. The margin over NMF on Replica is small (VOI 4.311 to 4.168, SC 0.188 to 0.209), so the claim of 'significantly better' performance is not statistically supported. Reporting variance and per-scene results, or including more scenes, is needed to substantiate the SOTA claim.
  4. [§3.3 and §3.5] The descriptors are constrained to unit norm (∥z∥=1), but the paper does not state how this constraint is enforced during gradient updates from L_seg or L_n. The closed-form least-squares solution in Eq. (6) is unconstrained and includes a bias term, and the mean-shift update in Eq. (8) relies on the von Mises-Fisher kernel, which assumes unit-norm descriptors. Without a renormalization step (e.g., projection onto the sphere after each optimizer step), the dot-product comparisons and the kernel become scale-dependent; the renormalization procedure must be specified.
minor comments (4)
  1. [Table 1] The column labeled 'Inference time' is misleading for optimization-based methods: the reported 16 minutes for PGS and 40 minutes for NMF are per-scene optimization/training times, not inference times in the usual sense. Renaming the column to something like 'Per-scene optimization time' would make the comparison clearer.
  2. [Contributions, line 1] The method is described as 'unsupervised,' but it relies on pretrained networks (SAM for masks and Omnidata for normals) to provide pseudo-labels. Consider using 'self-supervised' or 'without scene-level plane/depth labels' to avoid overstating the absence of learned priors.
  3. [Section 3.3 and Supplementary A] The thresholds for RAG edge cutting (10 cm planar distance, 20 degrees normal angle) are given only in the supplementary, and the GMT thresholds ϵ_B and ϵ_z are not reported anywhere. Including these thresholds in the main text, together with a sensitivity analysis, would improve reproducibility.
  4. [Eq. (6)] The matrix dimensions in Y = [Z|1]·W are not explicitly defined; clarifying that Z is the pixel-descriptor matrix, Y is the one-hot segment label matrix, and W includes the bias column would help readers follow the closed-form solution.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the plane-descriptor supervision depends on the field's geometry and external SAM/Omnidata cues, not on the descriptors being supervised, and the SOTA claim is checked against external ground-truth plane labels.

full rationale

The derivation chain is self-contained against external benchmarks. Plane descriptors z are optimized by Eq. 6 to predict merged SAM segment labels y. Those labels are produced by partitioning a Region Adjacency Graph using surface normals (Eq. 5, supervised by the external Omnidata normal predictions) and a planar distance dp (Eq. 7) that uses depth d(u,v) rendered from the Gaussian field. Crucially, the rendered depth and normals depend on Gaussian positions, opacities, and normal parameters, not on the descriptor vectors z being trained; labels are recomputed as geometry improves, but they are not defined in terms of the descriptor output or the final plane instances. The final 3D planes are obtained by probabilistic GMT grouping of Gaussians and are scored against ScanNet/Replica ground-truth plane labels, so the state-of-the-art claim is externally falsifiable. The only self-citation, NMF [46], is used as a comparison baseline rather than as a premise, and no uniqueness theorem or ansatz is imported by self-citation. The paper's admitted limitation about dark-region under-reconstruction is a robustness/correctness issue, not a circular dependency.

Assumptions & free parameters 8 free parameters · 5 assumptions · 1 invented entities

The central claim rests on a set of hand-chosen thresholds and weights, two external models (SAM and Omnidata), and a bootstrap from the Gaussian field's own rendered depth and normals. These are not free parameters in the physics sense, but they are fitted or tuned choices that the reported performance depends on. The most fragile dependence is the use of Gaussian-rendered depth in Equation (7), combined with the admitted under-reconstruction in dark areas.

free parameters (8)
  • Descriptor dimensionality k = 3
    The length of the plane descriptor vector is set to 3 without justification; it controls the capacity of the embedding and the sensitivity of the cosine-similarity merging in the Gaussian Mixture Tree (Section 3.3).
  • GMT merging thresholds epsilon_B and epsilon_z = not reported
    The tree merges nodes when Bhattacharya distance is below epsilon_B and descriptor dissimilarity below epsilon_z (Algorithm 1). The values are never given, so the grouping behavior is not reproducible and presumably was tuned.
  • RAG planar distance threshold = 10 cm
    Used in the Region Adjacency Graph to cut edges between SAM segments that likely belong to different planes (supplementary A).
  • RAG normal angle threshold = 20 degrees
    Used in the Region Adjacency Graph to cut edges when surface normals differ by more than this angle (supplementary A).
  • Mean-shift update rate eta = not reported
    Equation (8) applies a recurrent mean-shift update on descriptors with rate eta; the value of eta is not stated.
  • Mean-shift kernel bandwidth gamma = 60
    Bandwidth of the von Mises-Fisher kernel in the holistic separability step (supplementary A).
  • KNN neighborhood size K = 30
    Number of nearest neighbours used for local planar alignment and Laplacian smoothing of normals and descriptors (Section 3.4, supplementary A).
  • Loss weights for normal and segmentation terms = not reported
    The total training loss must combine the RGB loss (Eq. 2), normal loss (Eq. 5) and segmentation loss (Eq. 6) with weights; these weights are not stated, so the balance of supervision is unknown.
assumptions (5)
  • domain assumption SAM proposal masks, after merging, provide coherent groupings that overlap with true 3D planes.
    The descriptor learning (Section 3.3) treats refined SAM segments as labels for plane instances; if SAM oversegments or misses regions, the descriptors inherit the error.
  • domain assumption Omnidata predicts surface normals accurately enough to serve as supervision.
    Normal loss (Eq. 5) and RAG refinement (Section 3.3) rely on the off-the-shelf normal estimator [8]; erroneous normals directly corrupt the planar distance and edge cutting.
  • domain assumption The scene geometry is well captured by 3D Gaussian splatting after optimization, including in dark or textureless regions.
    The planar distance in Equation (7) uses depth rendered from the Gaussian field; the paper's own Limitations section admits dark regions suffer under-reconstruction, which can break the pseudo-label pipeline.
  • ad hoc to paper Bhattacharya distance between Gaussian distributions is a valid measure for deciding whether two groups of points lie on the same plane.
    The merging criterion in Algorithm 1 is a heuristic chosen by the authors; no derivation or comparison to alternative criteria is given.
  • ad hoc to paper The relationship between plane descriptors and segment membership is linear enough to be solved by closed-form linear regression.
    Equation (6) fits a linear map from descriptors to one-hot segment labels; if the descriptor manifold is nonlinear, this approximation limits the quality of the learned embeddings.
invented entities (1)
  • Plane descriptor vector z independent evidence
    purpose: A per-Gaussian latent embedding used to group Gaussians into distinct 3D plane instances in the Gaussian Mixture Tree.
    It is a learned internal representation; it has no measurement outside the method, but its quality is indirectly validated by the plane segmentation results against ground-truth labels on ScanNet and Replica.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Planar Gaussian Splatting." pith.science (2026). https://pith.science/paper/S5ZSHTXV

@misc{pith2026241201931,
  author       = {Pith},
  title        = {Pith review of: Planar Gaussian Splatting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/S5ZSHTXV}},
  note         = {Machine review of arXiv:2412.01931}
}
read the original abstract

This paper presents Planar Gaussian Splatting (PGS), a novel neural rendering approach to learn the 3D geometry and parse the 3D planes of a scene, directly from multiple RGB images. The PGS leverages Gaussian primitives to model the scene and employ a hierarchical Gaussian mixture approach to group them. Similar Gaussians are progressively merged probabilistically in the tree-structured Gaussian mixtures to identify distinct 3D plane instances and form the overall 3D scene geometry. In order to enable the grouping, the Gaussian primitives contain additional parameters, such as plane descriptors derived by lifting 2D masks from a general 2D segmentation model and surface normals. Experiments show that the proposed PGS achieves state-of-the-art performance in 3D planar reconstruction without requiring either 3D plane labels or depth supervision. In contrast to existing supervised methods that have limited generalizability and struggle under domain shift, PGS maintains its performance across datasets thanks to its neural rendering and scene-specific optimization mechanism, while also being significantly faster than existing optimization-based approaches.

Figures

Figures reproduced from arXiv: 2412.01931 by the authors.

Figure 1
Figure 1. The proposed PGS method constructs the entire scene using a tree-structured arrangement of Gaussian nodes. At the leaf [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Visualizing the steps of generating plane descriptors: [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Effect of holistic separation. (a) Ground-truth plane [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Examples of 3D planar reconstruction on ScanNet. (a) Ground-truth textured meshes (with holes on floor due to unseen regions [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Example of merging SAM proposals and lifting their boundary points into 3D. The boundary points of each segments is used for [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

48 extracted references · 40 canonical work pages

  1. [1]

    PlaneFormers: From sparse view planes to 3D re- construction

    Samir Agarwala, Linyi Jin, Chris Rockwell, and David F Fouhey. PlaneFormers: From sparse view planes to 3D re- construction. In Proceedings of the European Conference on Computer Vision, 2022. 1

    work page 2022

  2. [2]

    Controlling neural level sets

    Matan Atzmon, Niv Haim, Lior Yariv, Ofer Israelov, Haggai Maron, and Yaron Lipman. Controlling neural level sets. Advances in Neural Information Processing Systems , 2019. 1

    work page 2019

  3. [3]

    Contrastive lift: 3d object instance segmentation by slow-fast contrastive fusion

    Yash Bhalgat, Iro Laina, Jo ˜ao F Henriques, Andrew Zisser- man, and Andrea Vedaldi. Contrastive lift: 3d object instance segmentation by slow-fast contrastive fusion. arXiv preprint arXiv:2306.04633, 2023. 2, 3

    arXiv 2023

  4. [4]

    Seg- ment anything in 3d with nerfs

    Jiazhong Cen, Zanwei Zhou, Jiemin Fang, Wei Shen, Lingxi Xie, Dongsheng Jiang, Xiaopeng Zhang, Qi Tian, et al. Seg- ment anything in 3d with nerfs. Advances in Neural Infor- mation Processing Systems, 36, 2024. 2

    work page 2024

  5. [5]

    MobileNeRF: Exploiting the polygon ras- terization pipeline for efficient neural field rendering on mo- bile architectures

    Zhiqin Chen, Thomas Funkhouser, Peter Hedman, and An- drea Tagliasacchi. MobileNeRF: Exploiting the polygon ras- terization pipeline for efficient neural field rendering on mo- bile architectures. In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition, 2023. 1

    work page 2023

  6. [6]

    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 Proc. of the IEEE conf. on computer vision and pattern recognition, 2017. 6

    work page 2017

  7. [7]

    Depth-supervised NeRF: Fewer views and faster training for free

    Kangle Deng, Andrew Liu, Jun-Yan Zhu, and Deva Ra- manan. Depth-supervised NeRF: Fewer views and faster training for free. In Proceedings of the IEEE/CVF Conf. on Computer Vision and Pattern Recognition, 2022. 1

    work page 2022

  8. [8]

    Omnidata: A scalable pipeline for making multi- task mid-level vision datasets from 3D scans

    Ainaz Eftekhar, Alexander Sax, Jitendra Malik, and Amir Zamir. Omnidata: A scalable pipeline for making multi- task mid-level vision datasets from 3D scans. InProceedings of the IEEE/CVF International Conference on Computer Vi- sion, 2021. 4

    work page 2021

Show all 48 references
  1. [9]

    Nerf-sos: Any-view self- supervised object segmentation on complex scenes

    Zhiwen Fan, Peihao Wang, Yifan Jiang, Xinyu Gong, De- jia Xu, and Zhangyang Wang. Nerf-sos: Any-view self- supervised object segmentation on complex scenes. arXiv preprint arXiv:2209.08776, 2022. 2

    2022 arXiv

  2. [10]

    Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography.Communications of the ACM, 24(6):381–395, 1981

    Martin A Fischler and Robert C Bolles. Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography.Communications of the ACM, 24(6):381–395, 1981. 1, 6

    1981

  3. [11]

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

    Xiao Fu, Shangzhan Zhang, Tianrun Chen, Yichong Lu, Lanyun Zhu, Xiaowei Zhou, Andreas Geiger, and Yiyi Liao. Panoptic nerf: 3d-to-2d label transfer for panoptic urban scene segmentation. In 2022 International Conference on 3D Vision (3DV), pages 1–11. IEEE, 2022. 2

    2022

  4. [12]

    Learning deformable tetrahedral meshes for 3D reconstruction

    Jun Gao, Wenzheng Chen, Tommy Xiang, Alec Jacobson, Morgan McGuire, and Sanja Fidler. Learning deformable tetrahedral meshes for 3D reconstruction. Advances In Neu- ral Information Processing Systems, 33:9936–9947, 2020. 1

    2020

  5. [13]

    Neural 3D scene reconstruction with the manhattan-world assumption

    Haoyu Guo, Sida Peng, Haotong Lin, Qianqian Wang, Guofeng Zhang, Hujun Bao, and Xiaowei Zhou. Neural 3D scene reconstruction with the manhattan-world assumption. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022. 6

    2022

  6. [14]

    Methods for merging gaussian mixture components

    Christian Hennig. Methods for merging gaussian mixture components. Advances in data analysis and classification , 4:3–34, 2010. 4

    2010

  7. [15]

    Planar surface reconstruction from sparse views

    Linyi Jin, Shengyi Qian, Andrew Owens, and David F Fouhey. Planar surface reconstruction from sparse views. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021. 2

    2021

  8. [16]

    3d gaussian splatting for real-time radiance field rendering

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimk ¨uhler, and George Drettakis. 3d gaussian splatting for real-time radiance field rendering. ACM Transactions on Graphics , 42(4), 2023. 1, 3, 5, 6, 8, 11

    2023

  9. [17]

    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. arXiv preprint arXiv:2304.02643, 2023. 2, 4

    2023 arXiv

  10. [18]

    Recurrent pixel embed- ding for instance grouping

    Shu Kong and Charless C Fowlkes. Recurrent pixel embed- ding for instance grouping. In Proceedings of the IEEE con- ference on computer vision and pattern recognition , pages 9018–9028, 2018. 6

    2018

  11. [19]

    PlaneRCNN: 3D plane detection and reconstruc- tion from a single image

    Chen Liu, Kihwan Kim, Jinwei Gu, Yasutaka Furukawa, and Jan Kautz. PlaneRCNN: 3D plane detection and reconstruc- tion from a single image. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition ,

  12. [20]

    PlaneNet: Piece-wise planar reconstruc- tion from a single RGB image

    Chen Liu, Jimei Yang, Duygu Ceylan, Ersin Yumer, and Ya- sutaka Furukawa. PlaneNet: Piece-wise planar reconstruc- tion from a single RGB image. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition ,

  13. [21]

    PlaneMVS: 3D plane reconstruction from multi-view stereo

    Jiachen Liu, Pan Ji, Nitin Bansal, Changjiang Cai, Qingan Yan, Xiaolei Huang, and Yi Xu. PlaneMVS: 3D plane reconstruction from multi-view stereo. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022. 2

    2022

  14. [22]

    Instance neural radiance field

    Yichen Liu, Benran Hu, Junkai Huang, Yu-Wing Tai, and Chi-Keung Tang. Instance neural radiance field. In Proceed- ings of the IEEE/CVF International Conference on Com- puter Vision, pages 787–796, 2023. 2

    2023

  15. [23]

    Marching cubes: A high resolution 3D surface construction algorithm

    William E Lorensen and Harvey E Cline. Marching cubes: A high resolution 3D surface construction algorithm. ACM SIGGRAPH Computer Graphics, 21(4):163–169, 1987. 1

    1987

  16. [24]

    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 , 65(1):99–106, 2021. 1

    2021

  17. [25]

    Laterf: Label and text driven object radiance fields

    Ashkan Mirzaei, Yash Kant, Jonathan Kelly, and Igor Gilitschenski. Laterf: Label and text driven object radiance fields. In European Conference on Computer Vision, pages 20–36. Springer, 2022. 2

    2022

  18. [26]

    Extracting triangular 3D models, materials, and lighting from images

    Jacob Munkberg, Jon Hasselgren, Tianchang Shen, Jun Gao, Wenzheng Chen, Alex Evans, Thomas M¨uller, and Sanja Fi- dler. Extracting triangular 3D models, materials, and lighting from images. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022. 1

    2022

  19. [27]

    Atlas: End- to-end 3d scene reconstruction from posed images

    Zak Murez, Tarrence Van As, James Bartolozzi, Ayan Sinha, Vijay Badrinarayanan, and Andrew Rabinovich. Atlas: End- to-end 3d scene reconstruction from posed images. In Pro- ceedings of the European Conference on Computer Vision , pages 414–431, 2020. 12

    2020

  20. [28]

    Learning pairwise inter-plane relations for piecewise planar reconstruction

    Yiming Qian and Yasutaka Furukawa. Learning pairwise inter-plane relations for piecewise planar reconstruction. In Proceedings of the European Conference on Computer Vi- sion, 2020. 2

    2020

  21. [29]

    Neural groundplans: Persistent neural scene representations from a single image

    Prafull Sharma, Ayush Tewari, Yilun Du, Sergey Zakharov, Rares Andrei Ambrus, Adrien Gaidon, William T Freeman, Fr´edo Durand, Joshua B Tenenbaum, and Vincent Sitzmann. Neural groundplans: Persistent neural scene representations from a single image. In International Conference...

    2023

  22. [30]

    Deep marching tetrahedra: a hybrid represen- tation for high-resolution 3D shape synthesis

    Tianchang Shen, Jun Gao, Kangxue Yin, Ming-Yu Liu, and Sanja Fidler. Deep marching tetrahedra: a hybrid represen- tation for high-resolution 3D shape synthesis. Advances in Neural Information Processing Systems, 2021. 1

    2021

  23. [31]

    Panoptic lifting for 3d scene understanding with neural fields

    Yawar Siddiqui, Lorenzo Porzi, Samuel Rota Bul `o, Nor- man M ¨uller, Matthias Nießner, Angela Dai, and Peter Kontschieder. Panoptic lifting for 3d scene understanding with neural fields. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, p...

    2023

  24. [32]

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

    1906 arXiv

  25. [33]

    NeuralRecon: Real-time coherent 3D recon- struction from monocular video

    Jiaming Sun, Yiming Xie, Linghao Chen, Xiaowei Zhou, and Hujun Bao. NeuralRecon: Real-time coherent 3D recon- struction from monocular video. In Proc. of the IEEE/CVF Conf. on Computer Vision and Pattern Recognition , pages 15598–15607, 2021. 6, 8

    2021

  26. [34]

    PlaneTR: Structure-guided transformers for 3d plane recov- ery

    Bin Tan, Nan Xue, Song Bai, Tianfu Wu, and Gui-Song Xia. PlaneTR: Structure-guided transformers for 3d plane recov- ery. In Proceedings of the IEEE/CVF International Confer- ence on Computer Vision, 2021. 1, 2, 6

    2021

  27. [35]

    Neu- raldiff: Segmenting 3d objects that move in egocentric videos

    Vadim Tschernezki, Diane Larlus, and Andrea Vedaldi. Neu- raldiff: Segmenting 3d objects that move in egocentric videos. In 2021 International Conference on 3D Vision (3DV), pages 910–919. IEEE, 2021. 2

    2021

  28. [36]

    Barron, and Pratul P

    Dor Verbin, Peter Hedman, Ben Mildenhall, Todd Zickler, Jonathan T. Barron, and Pratul P. Srinivasan. Ref-NeRF: Structured view-dependent appearance for neural radiance fields. Proceedings of the IEEE/CVF International Confer- ence on Computer Vision, 2022. 1

    2022

  29. [37]

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

    Suhani V ora, Noha Radwan, Klaus Greff, Henning Meyer, Kyle Genova, Mehdi SM Sajjadi, Etienne Pot, Andrea Tagliasacchi, and Daniel Duckworth. Nesf: Neural semantic fields for generalizable semantic segmentation of 3d scenes. arXiv preprint arXiv:2111.13260, 2021. 2

    2021 arXiv

  30. [38]

    Neus: Learning neural implicit surfaces by volume rendering for multi-view reconstruction

    Peng Wang, Lingjie Liu, Yuan Liu, Christian Theobalt, Taku Komura, and Wenping Wang. Neus: Learning neural implicit surfaces by volume rendering for multi-view reconstruction. arXiv preprint arXiv:2106.10689, 2021. 1

    2021 arXiv

  31. [39]

    Fig-nerf: Figure-ground neural radi- ance fields for 3d object category modelling

    Christopher Xie, Keunhong Park, Ricardo Martin-Brualla, and Matthew Brown. Fig-nerf: Figure-ground neural radi- ance fields for 3d object category modelling. In 2021 In- ternational Conference on 3D Vision (3DV), pages 962–971. IEEE, 2021. 2

    2021

  32. [40]

    PlanarRecon: Real-time 3D plane detection and reconstruction from posed monocular videos

    Yiming Xie, Matheus Gadelha, Fengting Yang, Xiaowei Zhou, and Huaizu Jiang. PlanarRecon: Real-time 3D plane detection and reconstruction from posed monocular videos. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022. 1, 2, 6, 7, 8, 12

    2022

  33. [41]

    Recovering 3D planes from a single image via convolutional neural networks

    Fengting Yang and Zihan Zhou. Recovering 3D planes from a single image via convolutional neural networks. In Pro- ceedings of the European Conference on Computer Vision (ECCV), 2018. 2

    2018

  34. [42]

    V ol- ume rendering of neural implicit surfaces

    Lior Yariv, Jiatao Gu, Yoni Kasten, and Yaron Lipman. V ol- ume rendering of neural implicit surfaces. Advances in Neu- ral Information Processing Systems, 34:4805–4815, 2021. 1

    2021

  35. [43]

    Unsu- pervised discovery of object radiance fields

    Hong-Xing Yu, Leonidas J Guibas, and Jiajun Wu. Unsu- pervised discovery of object radiance fields. arXiv preprint arXiv:2107.07905, 2021. 2

    2021 arXiv

  36. [44]

    MonoSDF: Exploring monocu- lar geometric cues for neural implicit surface reconstruction

    Zehao Yu, Songyou Peng, Michael Niemeyer, Torsten Sat- tler, and Andreas Geiger. MonoSDF: Exploring monocu- lar geometric cues for neural implicit surface reconstruction. Advances in neural information processing systems, 2022. 1

    2022

  37. [45]

    Single-image piece-wise planar 3d recon- struction via associative embedding

    Zehao Yu, Jia Zheng, Dongze Lian, Zihan Zhou, and Shenghua Gao. Single-image piece-wise planar 3d recon- struction via associative embedding. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019. 1, 2

    2019

  38. [46]

    Neural mesh fusion: Unsu- pervised 3d planar surface understanding

    Farhad G Zanjani, Hong Cai, Yinhao Zhu, Leyla Mir- vakhabova, and Fatih Porikli. Neural mesh fusion: Unsu- pervised 3d planar surface understanding. arXiv preprint arXiv:2402.16739, 2024. 1, 2, 6, 8

    2024 arXiv

  39. [47]

    In-place scene labelling and understanding with implicit scene representation

    Shuaifeng Zhi, Tristan Laidlow, Stefan Leutenegger, and An- drew J Davison. In-place scene labelling and understanding with implicit scene representation. In Proceedings of the IEEE/CVF International Conference on Computer Vision ,

  40. [2021]

    2 Planar Gaussian Splatting — Supplementary Material — Farhad G. Zanjani Hong Cai Hanno Ackermann Leila Mirvakhabova Fatih Porikli Qualcomm AI Research1 {fzanjani, hongcai, hackerma, lmirvakh, fporikli }@qti.qualcomm.com This supplementary document includes further implementat...

Pith tools

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