Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

See In Detail: Enhancing Sparse-view 3D Gaussian Splatting with Local Depth and Semantic Regularization

T0 review · 3 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read This paper claims that sparse-view 3D Gaussian Splatting can be substantially improved by adding a DINO-ViT semantic-consistency loss and a DPT-based local depth loss, outperforming prior methods by up to 0.4 dB PSNR on LLFF.

desk verdict Equation (5) makes the local depth regularizer push correlation to zero instead of +1, contradicting the paper's own description and undermining one of its two central contributions; the semantic regularization is reasonable and the paper is salvageable, but not in current form. read the letter →

arxiv 2501.11508 v1 pith:UVQSBS4G submitted 2025-01-20 cs.CV

Zongqi He , Zhe Xiao , Kin-Chung Chan , Yushen Zuo , Jun Xiao , Kin-Man Lam This is my paper · ORCID
classification cs.CV
keywords Sparse-viewnovelviewsynthesis3DGaussianSplattingSemanticregularizationLocaldepthDINO-ViTfeaturesDPTpriorsLLFFdatasetMulti-viewconsistency
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

Sparse-view 3D Gaussian Splatting (3DGS) produces distorted geometry and lost detail when only a handful of images are available. This paper proposes a method, called SIDGaussian, that adds two prior-based losses to the standard sparse-view 3DGS training loop: a semantic consistency loss that aligns DINO-ViT features between rendered training views and rendered side views, and a local depth loss that matches rendered side-view depth patches to DPT monocular depth estimates in a scale-invariant way. The aim is to make sparse-input 3DGS competitive with NeRF-based sparse-view methods while retaining real-time rendering. On the LLFF dataset with three training views, the reported result is a PSNR of 20.71 dB at 1/8 resolution, a gain of 0.4 dB over the strongest prior method, together with the lowest LPIPS among all compared methods.

What carries the argument

Two auxiliary loss terms grafted onto the FSGS-style objective. The semantic regularizer uses a pretrained DINO-ViT encoder as a fixed perceptual teacher: it embeds randomly cropped patches from rendered training views and rendered side views and minimizes their L2 distance, so side views are pulled toward the semantic content of the training views. The depth regularizer uses a pretrained Dense Prediction Transformer (DPT) as a monocular depth teacher; it locally normalizes rendered and predicted depth patches with $d_{\mathrm{LN}}(x) = (d(x)-\mu)/(\sigma+\epsilon)$ and measures their similarity with Pearson correlation, making the geometric constraint invariant to depth scale and offset while still sensitive to local shape. The two losses are combined with the base color, SSIM, and global depth loss in a weighted sum, so the regularizers act directly on rendered images and require no extra data at inference time.

What would settle it

Render side views from the trained model, compute the local Pearson correlation between the model's rendered depth and DPT's depth on those same views, and check whether the scenes with the largest PSNR gains also have the highest correlations; if improved scenes show near-zero correlation, the depth term is not doing the geometric work claimed.

Watch

Extended reading notes

Core claim

The central claim is that an under-constrained 3DGS scene can be stabilized by supervising two things the standard loss never sees: semantic agreement across views and local relative depth structure. The semantic loss $L_{\mathrm{sem}}$ computes the L2 distance between DINO-ViT embeddings of randomly cropped patches from training-view and side-view renderings, encouraging unseen views to carry the same content as the observed views. The local depth loss $L_{\mathrm{depth}}$ renders depth for side views, normalizes each local 126x126 patch by its own mean and standard deviation, and uses Pearson correlation to compare the normalized patches with DPT predictions. Because both patch normalization and Pearson correlation remove scale and shift, the depth term constrains relative structure rather than absolute metric depth. In Table I the full method reports PSNR 20.71 / SSIM 0.708 / LPIPS 0.205 at 1/8 resolution, and the ablation in Table II shows semantic regularization alone lifts PSNR from 20.31 to 20.57 while the addition of local depth reaches 20.71.

Load-bearing premise

The load-bearing premise is that DPT monocular depth for the rendered side views is accurate enough to supervise geometry even though the 3DGS model generating those side views is still being trained; if DPT is systematically biased on those synthetic views, the local depth loss would reinforce wrong geometry rather than correct it.

Editorial extensions

If this is right

  • Three-view 3DGS becomes a more viable real-time alternative to sparse-view NeRF methods on forward-facing scenes, since the extra losses add no inference-time cost.
  • The semantic term alone accounts for most of the perceptual improvement in the ablation (LPIPS drops from 0.288 to 0.220), suggesting that enforcing semantic consistency across viewpoints is a strong prior for ill-posed radiance-field optimization.
  • Local depth regularization adds a smaller but consistent gain on top of the semantic term (PSNR from 20.57 to 20.71), indicating that scale-invariant depth matching captures detail that global depth supervision misses.
  • The total loss remains a simple weighted sum, so each regularizer can be switched on or off per scene while the base 3DGS pipeline stays unchanged.

Reading between the lines

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

  • The training loop is effectively a self-distillation scheme in which DINO and DPT act as fixed teachers for views the model has never seen; a natural stress test is two-view or single-view input, where the priors are weaker but matter more.
  • Because the depth term is invariant to per-patch scale and shift, it cannot correct global depth drift or metric scale errors; pairing it with an ordinal or ranking-based depth loss could close that gap.
  • The gains are reported only on LLFF's forward-facing scenes; unbounded or object-centric captures would test whether the semantic and depth priors survive large scale variation and occlusions.
  • The semantic loss's reliance on random patch crops suggests a simple extension: multi-scale crops or attention-weighted feature maps from several DINO layers could strengthen the consistency signal.
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

3 major / 5 minor

Summary. The manuscript presents SIDGaussian, a sparse-view 3D Gaussian Splatting method that augments the FSGS baseline with two regularizers. A semantic regularization term (Eq. (3)) penalizes the L2 distance between DINO-ViT features of patches from rendered training and side views, and a local depth regularization term (Eq. (5)) is defined as the L1 norm of the Pearson correlation between locally normalized DPT depth patches and rendered depth patches. Experiments on LLFF with three training views report state-of-the-art PSNR/SSIM/LPIPS at 1/8 and 1/4 resolution, with an ablation in Table II showing gains from both semantic and depth terms. The paper claims up to 0.4 dB PSNR improvement over FSGS.

Significance. If the contributions were correctly formulated, SIDGaussian would be a modest but useful step for real-time sparse-view rendering: the semantic prior is principled, the comparison is direct, and held-out-view evaluation with external baselines avoids circularity. The reported numerical gains are positive across all three metrics. However, the principal geometric prior is written incorrectly, and the DINO-ViT citation is wrong; both must be corrected before the empirical claims can be interpreted. With the current Eq. (5), the depth term does not implement the described similarity objective, so the ablation's depth row cannot be credited to the stated mechanism. The paper is not acceptable in this form.

major comments (3)
  1. [II-B2, Eq. (5)] Equation (5) defines L_depth = ||Corr(d_LN, d_hat_LN)||_1, with Corr the Pearson correlation. Since Corr lies in [-1,1], minimizing this absolute value drives the correlation toward 0, not toward +1. The text states that the loss "encourages the similarity between local patches" and uses correlation "to measure the similarity"; that objective would require minimizing 1 - Corr or -Corr. Under the equation as written, the gradient prefers zero correlation (or a zero-variance rendered patch), which is the opposite of aligning the rendered depth structure with the DPT prediction. Consequently, the Table II ablation row labeled "local depth regularization" does not demonstrate the claimed mechanism. This is the central load-bearing issue for the depth contribution.
  2. [II-B1, reference [40]] The semantic regularizer is claimed to use "DINO-ViT [40]", but reference [40] is "DINO: Detr with improved denoising anchor boxes for end-to-end object detection" (arXiv:2203.03605), not the self-supervised DINO ViT feature extractor of Caron et al. The paper must cite the correct pretrained model and specify which layer and patchification are used; as written, the semantic regularizer is not reproducible from the bibliography.
  3. [II-B2 and III-A] The depth prior assumes DPT predictions for rendered side views are reliable. These side views come from an under-constrained 3DGS model that is itself being optimized, and DPT is not trained on synthetic rendered images. The paper provides no validation of this assumption, e.g., by comparing DPT side-view depth against ground-truth depth on held-out views or by ablating the choice of side-view source. Even after Eq. (5) is corrected, this correctness risk remains and should be analyzed.
minor comments (5)
  1. [III-A] The loss weights omega_0, omega_sem, and omega_depth are not reported; only the patch size and iteration count are given. Specify all hyperparameters for reproducibility.
  2. [II-A, Eq. (2)] The function d(.) in Eq. (2) is called a "distance function" but is never defined, while d_LN and d_hat_LN are later used for depth maps; the notation should be disambiguated.
  3. [II-B2, Eq. (4)] The symbol P is first used as a set of patches (P ⊆ {Pr, Pt}) and then as the spatial support of x in the normalization formula; this overloaded notation should be clarified.
  4. [References] References [43] and [48] are the same paper and should be merged; the abstract also contains the typo "inphut".
  5. [III-B and Fig. 5] The paper claims real-time rendering but reports no runtime; Fig. 5 also does not specify the axes or the exact weight values used, so the sensitivity discussion is difficult to reproduce.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's central claims are evaluated on held-out LLFF views against external baselines, and the self-citations are motivational rather than load-bearing.

full rationale

The sparse-view rendering claims are supported by experiments on held-out views of the LLFF dataset and comparisons with external methods; no parameter is fitted to the target test views and then reported as a prediction. The authors' self-citation [47] is used only to motivate the Pearson-correlation choice as a soft metric for scale-ambiguous depth patches, and the cited prior work does not supply a uniqueness theorem or a result that forces the paper's central derivation. References [43], [48] are similarly invoked only to note that semantic regularization has been effective elsewhere, which is an ordinary literature motivation, not load-bearing. The local depth and semantic regularization losses are part of the training objective, and their claimed contribution is verified by ablations, so there is no reduction of a reported result to a fitted input. The reviewer's main technical concern is that Eq. (5) minimizes the absolute value of the Pearson correlation, which would drive correlation toward zero rather than toward positive similarity; however, that is a correctness or sign-error issue in the loss design, not a circularity in the derivation chain. The paper is therefore self-contained against external benchmarks, and the self-citations do not elevate the circularity score.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The central claim rests on three domain assumptions about pretrained models (DINO and DPT) and about the meaning of correlation between normalized depth patches. The free parameters are the loss weights and patch size, all chosen on the LLFF benchmark. No new physical or geometric entities are introduced.

free parameters (5)
  • omega_0 = not reported
    Weight of the base FSGS loss in Eq. (7), inherited from [39] and not specified. It affects the balance between photometric and depth losses.
  • omega_sem = not reported (Fig. 5 varies it)
    Weight of the semantic regularization, tuned on LLFF. The sensitivity study in Fig. 5 shows that lower values cause greater PSNR fluctuation.
  • omega_depth = not reported
    Weight of the local depth regularization, tuned on LLFF. Fig. 5 suggests higher values correlate with higher PSNR.
  • local patch size = 126x126
    Patch size for local depth normalization, chosen by hand in Section III-A. This directly controls the scale of the depth correlation.
  • number of iterations = 1.2e4
    Fixed iteration count in Section III-A. Not tuned per scene, but a chosen hyperparameter.
assumptions (3)
  • domain assumption DINO-ViT features are semantically consistent across views of the same scene.
    Used to define Lsem in Eq. (3) without validation on the target scenes. The paper relies on the pretrained model to align training and side views.
  • domain assumption DPT monocular depth maps provide reliable local geometry for sparse-view scenes, including rendered side views.
    Used as pseudo ground truth dLN in Eq. (5). The accuracy of DPT on synthetic rendered side views is not assessed.
  • domain assumption Pearson correlation after local normalization measures geometric similarity of depth patches.
    Used in Eq. (6). Assumes scale and shift invariance are sufficient for good geometry, which is plausible but not proven.

how reviews work

0 comments
Cite this review

Pith. "Pith review of See In Detail: Enhancing Sparse-view 3D Gaussian Splatting with Local Depth and Semantic Regularization." pith.science (2026). https://pith.science/paper/UVQSBS4G

@misc{pith2026250111508,
  author       = {Pith},
  title        = {Pith review of: See In Detail: Enhancing Sparse-view 3D Gaussian Splatting with Local Depth and Semantic Regularization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UVQSBS4G}},
  note         = {Machine review of arXiv:2501.11508}
}
read the original abstract

3D Gaussian Splatting (3DGS) has shown remarkable performance in novel view synthesis. However, its rendering quality deteriorates with sparse inphut views, leading to distorted content and reduced details. This limitation hinders its practical application. To address this issue, we propose a sparse-view 3DGS method. Given the inherently ill-posed nature of sparse-view rendering, incorporating prior information is crucial. We propose a semantic regularization technique, using features extracted from the pretrained DINO-ViT model, to ensure multi-view semantic consistency. Additionally, we propose local depth regularization, which constrains depth values to improve generalization on unseen views. Our method outperforms state-of-the-art novel view synthesis approaches, achieving up to 0.4dB improvement in terms of PSNR on the LLFF dataset, with reduced distortion and enhanced visual quality.

Figures

Figures reproduced from arXiv: 2501.11508 by the authors.

Figure 1
Figure 1. Visual results of FreeNeRF [38], RegNeRF [33], FSGS [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The overall pipeline of our proposed SIDGaussian. A sparse point cloud for the 3D Gaussian initialization is generated [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Visual results of the scenes “Leaves” and “Horns” generated by FreeNeRF [38], RegNeRF [33], FSGS [39], and our [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Visual results of our method with/without semantic [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: The influence of the weights wsem and wdepth in terms of PSNR score. IV. CONCLUSION In this paper, we focus on novel view synthesis based on sparse inputs and propose a sparse-view 3D Gaussian splatting model, namely SIDGaussian. To ensure multi-view consistency, we pr…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Multi-view Normal and Distance Guidance Gaussian Splatting for Surface Reconstruction

    cs.CV 2025-08 unverdicted novelty 4.0 of 10

    A 3DGS surface reconstruction method that enforces multi-view distance and normal consistency between nearby views to reduce geometry drift.

Reference graph

Works this paper leans on

49 extracted references · 40 canonical work pages · cited by 1 Pith paper

  1. [40]

    Dino: Detr with improved denoising anchor boxes for end-to- end object detection,

    H. Zhang, F. Li, S. Liu, L. Zhang, H. Su, J. Zhu, L. M. Ni, and H.-Y . Shum, “Dino: Detr with improved denoising anchor boxes for end-to- end object detection,” arXiv preprint arXiv:2203.03605 , 2022

    arXiv 2022

  2. [1]

    A survey on vision-based human action recognition,

    R. Poppe, “A survey on vision-based human action recognition,” Image and vision computing , vol. 28, no. 6, pp. 976–990, 2010

    work page 2010

  3. [2]

    Online video super-resolution with convolutional kernel bypass grafts,

    J. Xiao, X. Jiang, N. Zheng, H. Yang, Y . Yang, Y . Yang, D. Li, and K.-M. Lam, “Online video super-resolution with convolutional kernel bypass grafts,” IEEE TMM , vol. 25, pp. 8972–8987, 2023

    work page 2023

  4. [3]

    3-d model-based tracking of humans in action: a multi-view approach,

    D. M. Gavrila and L. S. Davis, “3-d model-based tracking of humans in action: a multi-view approach,” in CVPR, vol. 96, p. 73, 1996

    work page 1996

  5. [4]

    Towards Multi-View Consistent Style Transfer with One-Step Diffusion via Vision Conditioning

    Y . Zuo, J. Xiao, K.-C. Chan, R. Dong, C. Yang, Z. He, H. Xie, and K.-M. Lam, “Towards multi-view consistent style transfer with one- step diffusion via vision conditioning,”arXiv preprint arXiv:2411.10130, 2024

    work page Pith review arXiv 2024

  6. [5]

    Cross-view action recognition from temporal self-similarities,

    I. N. Junejo, E. Dexter, I. Laptev, and P. P ´erez, “Cross-view action recognition from temporal self-similarities,” in ECCV, pp. 293–306, Springer, 2008

    work page 2008

  7. [6]

    Inte- grally mixing pyramid representations for anchor-free object detection in aerial imagery,

    C. Zhang, J. Xiao, C. Yang, J. Zhou, K.-M. Lam, and Q. Wang, “Inte- grally mixing pyramid representations for anchor-free object detection in aerial imagery,” IEEE Geoscience and Remote Sensing Letters , 2024

    work page 2024

  8. [7]

    A comparison and evaluation of multi-view stereo reconstruction algo- rithms,

    S. M. Seitz, B. Curless, J. Diebel, D. Scharstein, and R. Szeliski, “A comparison and evaluation of multi-view stereo reconstruction algo- rithms,” in CVPR, vol. 1, pp. 519–528, IEEE, 2006

    work page 2006

Show all 49 references
  1. [8]

    Accurate, dense, and robust multiview stereopsis,

    Y . Furukawa and J. Ponce, “Accurate, dense, and robust multiview stereopsis,” IEEE TPAMI, vol. 32, no. 8, pp. 1362–1376, 2009

    2009

  2. [9]

    3d-r2n2: A unified approach for single and multi-view 3d object reconstruction,

    C. B. Choy, D. Xu, J. Gwak, K. Chen, and S. Savarese, “3d-r2n2: A unified approach for single and multi-view 3d object reconstruction,” in ECCV, pp. 628–644, Springer, 2016

    2016

  3. [10]

    Structured adversarial self-supervised learning for robust object detection in remote sensing images,

    C. Zhang, K.-M. Lam, T. Liu, Y .-L. Chan, and Q. Wang, “Structured adversarial self-supervised learning for robust object detection in remote sensing images,” IEEE TGRS , 2024

    2024

  4. [11]

    A point set generation network for 3d object reconstruction from a single image,

    H. Fan, H. Su, and L. J. Guibas, “A point set generation network for 3d object reconstruction from a single image,” in CVPR, pp. 605–613, 2017

    2017

  5. [12]

    Feature redundancy mining: Deep light-weight image super-resolution model,

    J. Xiao, W. Jia, and K.-M. Lam, “Feature redundancy mining: Deep light-weight image super-resolution model,” in ICASSP, pp. 1620–1624, IEEE, 2021

    2021

  6. [13]

    Hierarchical vertex-wise intensification graph convolution for skeleton-based activity recognition,

    Y . Li, H. Xie, J. Xiao, C. Zhang, T. Liu, and K.-M. Lam, “Hierarchical vertex-wise intensification graph convolution for skeleton-based activity recognition,” in ICIP, pp. 2702–2708, IEEE, 2024

    2024

  7. [14]

    Towards pro- gressive multi-frequency representation for image warping,

    J. Xiao, Z. Lyu, C. Zhang, Y . Ju, C. Shui, and K.-M. Lam, “Towards pro- gressive multi-frequency representation for image warping,” in CVPR, pp. 2995–3004, 2024

    2024

  8. [15]

    Image-based 3d modelling: a review,

    F. Remondino and S. El-Hakim, “Image-based 3d modelling: a review,” The photogrammetric record , vol. 21, no. 115, pp. 269–291, 2006

    2006

  9. [16]

    Self-feature learning: An efficient deep lightweight network for image super-resolution,

    J. Xiao, Q. Ye, R. Zhao, K.-M. Lam, and K. Wan, “Self-feature learning: An efficient deep lightweight network for image super-resolution,” in ACM MM, pp. 4408–4416, 2021

    2021

  10. [17]

    Pointshop 3d: An interactive system for point-based surface editing,

    M. Zwicker, M. Pauly, O. Knoll, and M. Gross, “Pointshop 3d: An interactive system for point-based surface editing,” ACM TOG, vol. 21, no. 3, pp. 322–329, 2002

    2002

  11. [18]

    Deep progressive fea- ture aggregation network for multi-frame high dynamic range imaging,

    J. Xiao, Q. Ye, T. Liu, C. Zhang, and K.-M. Lam, “Deep progressive fea- ture aggregation network for multi-frame high dynamic range imaging,” Neurocomputing, vol. 594, p. 127804, 2024

    2024

  12. [19]

    Shape completion enabled robotic grasping,

    J. Varley, C. DeChant, A. Richardson, J. Ruales, and P. Allen, “Shape completion enabled robotic grasping,” in IEEE/RSJ IROS , pp. 2442– 2447, IEEE, 2017

    2017

  13. [20]

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

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

    2021

  14. [21]

    Mip-nerf: A multiscale representation for anti- aliasing neural radiance fields,

    J. T. Barron, B. Mildenhall, M. Tancik, P. Hedman, R. Martin-Brualla, and P. P. Srinivasan, “Mip-nerf: A multiscale representation for anti- aliasing neural radiance fields,” in ICCV, pp. 5855–5864, 2021

    2021

  15. [22]

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

    J. T. Barron, B. Mildenhall, D. Verbin, P. P. Srinivasan, and P. Hedman, “Mip-nerf 360: Unbounded anti-aliased neural radiance fields,” inCVPR, pp. 5470–5479, 2022

    2022

  16. [23]

    Nerf++: Analyzing and improving neural radiance fields,

    K. Zhang, G. Riegler, N. Snavely, and V . Koltun, “Nerf++: Analyzing and improving neural radiance fields,” arXiv preprint arXiv:2010.07492, 2020

    2010 arXiv

  17. [24]

    Point-nerf: Point-based neural radiance fields,

    Q. Xu, Z. Xu, J. Philip, S. Bi, Z. Shu, K. Sunkavalli, and U. Neumann, “Point-nerf: Point-based neural radiance fields,” in CVPR, pp. 5438– 5448, 2022

    2022

  18. [25]

    Neural sparse voxel fields,

    L. Liu, J. Gu, K. Zaw Lin, T.-S. Chua, and C. Theobalt, “Neural sparse voxel fields,” Advances in Neural Information Processing Systems , vol. 33, pp. 15651–15663, 2020

    2020

  19. [26]

    Light field networks: Neural scene representations with single- evaluation rendering,

    V . Sitzmann, S. Rezchikov, B. Freeman, J. Tenenbaum, and F. Du- rand, “Light field networks: Neural scene representations with single- evaluation rendering,” Advances in Neural Information Processing Sys- tems, vol. 34, pp. 19313–19325, 2021

    2021

  20. [27]

    3d gaussian splatting for real-time radiance field rendering,

    B. Kerbl, G. Kopanas, T. Leimk ¨uhler, and G. Drettakis, “3d gaussian splatting for real-time radiance field rendering,” ACM TOG, vol. 42, July 2023

    2023

  21. [28]

    Luciddreamer: Domain-free generation of 3d gaussian splatting scenes,

    J. Chung, S. Lee, H. Nam, J. Lee, and K. M. Lee, “Luciddreamer: Domain-free generation of 3d gaussian splatting scenes,” arXiv preprint arXiv:2311.13384, 2023

    2023 arXiv

  22. [29]

    Multi-scale 3d gaussian splatting for anti-aliased rendering,

    Z. Yan, W. F. Low, Y . Chen, and G. H. Lee, “Multi-scale 3d gaussian splatting for anti-aliased rendering,” in CVPR, pp. 20923–20931, 2024

    2024

  23. [30]

    Pixel-gs: Density control with pixel-aware gradient for 3d gaussian splatting,

    Z. Zhang, W. Hu, Y . Lao, T. He, and H. Zhao, “Pixel-gs: Density control with pixel-aware gradient for 3d gaussian splatting,” arXiv preprint arXiv:2403.15530, 2024

    2024 arXiv

  24. [31]

    Compact 3d gaussian representation for radiance field,

    J. C. Lee, D. Rho, X. Sun, J. H. Ko, and E. Park, “Compact 3d gaussian representation for radiance field,” in CVPR, pp. 21719–21728, 2024

    2024

  25. [32]

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

    T. Lu, M. Yu, L. Xu, Y . Xiangli, L. Wang, D. Lin, and B. Dai, “Scaffold- gs: Structured 3d gaussians for view-adaptive rendering,” in CVPR, pp. 20654–20664, 2024

    2024

  26. [33]

    Regnerf: Regularizing neural radiance fields for view synthesis from sparse inputs,

    M. Niemeyer, J. T. Barron, B. Mildenhall, M. S. Sajjadi, A. Geiger, and N. Radwan, “Regnerf: Regularizing neural radiance fields for view synthesis from sparse inputs,” in CVPR, pp. 5480–5490, 2022

    2022

  27. [34]

    Putting nerf on a diet: Semantically consistent few-shot view synthesis,

    A. Jain, M. Tancik, and P. Abbeel, “Putting nerf on a diet: Semantically consistent few-shot view synthesis,” in ICCV, pp. 5885–5894, 2021

    2021

  28. [35]

    Learning transferable visual models from natural language supervision,

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, et al., “Learning transferable visual models from natural language supervision,” in International conference on machine learning , pp. 8748–8763, PMLR, 2021

    2021

  29. [36]

    Sparsenerf: Distilling depth ranking for few-shot novel view synthesis,

    G. Wang, Z. Chen, C. C. Loy, and Z. Liu, “Sparsenerf: Distilling depth ranking for few-shot novel view synthesis,” in ICCV, pp. 9065–9076, 2023

    2023

  30. [37]

    Vision transformers for dense prediction,

    R. Ranftl, A. Bochkovskiy, and V . Koltun, “Vision transformers for dense prediction,” in ICCV, pp. 12179–12188, 2021

    2021

  31. [38]

    Freenerf: Improving few-shot neural rendering with free frequency regularization,

    J. Yang, M. Pavone, and Y . Wang, “Freenerf: Improving few-shot neural rendering with free frequency regularization,” in CVPR, 2023

    2023

  32. [39]

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

    Z. Zhu, Z. Fan, Y . Jiang, and Z. Wang, “Fsgs: Real-time few-shot view synthesis using gaussian splatting,” 2023

    2023

  33. [41]

    Dense depth priors for neural radiance fields from sparse input views,

    B. Roessle, J. T. Barron, B. Mildenhall, P. P. Srinivasan, and M. Nießner, “Dense depth priors for neural radiance fields from sparse input views,” in CVPR, pp. 12892–12901, 2022

    2022

  34. [42]

    Pearson correlation coefficient,

    I. Cohen, Y . Huang, J. Chen, J. Benesty, J. Benesty, J. Chen, Y . Huang, and I. Cohen, “Pearson correlation coefficient,” Noise reduction in speech processing, pp. 1–4, 2009

    2009

  35. [44]

    A structure-affinity dual attention-based network to segment spine for scoliosis assessment,

    H. Xie, Z. Huang, F. H. Leung, Y . Ju, Y .-P. Zheng, and S. H. Ling, “A structure-affinity dual attention-based network to segment spine for scoliosis assessment,” in BIBM, pp. 1567–1574, IEEE, 2023

    2023

  36. [45]

    Sinnerf: Training neural radiance fields on complex scenes from a single image,

    D. Xu, Y . Jiang, P. Wang, Z. Fan, H. Shi, and Z. Wang, “Sinnerf: Training neural radiance fields on complex scenes from a single image,” in ECCV, pp. 736–753, Springer, 2022

    2022

  37. [46]

    Satr: A structure-affinity attention-based transformer encoder for spine segmentation,

    H. Xie, Z. Huang, F. H. Leung, N. Law, Y . Ju, Y .-P. Zheng, and S. H. Ling, “Satr: A structure-affinity attention-based transformer encoder for spine segmentation,” in ISBI, pp. 1–5, IEEE, 2024

    2024

  38. [47]

    Point cloud densification for 3d gaussian splatting from sparse input views,

    K.-C. Chan, J. Xiao, H. L. Goshu, and K.-m. Lam, “Point cloud densification for 3d gaussian splatting from sparse input views,” in ACM MM

  39. [48]

    Deep multi-scale feature mixture model for image super-resolution with multiple-focal- length degradation,

    J. Xiao, Q. Ye, R. Zhao, K.-M. Lam, and K. Wan, “Deep multi-scale feature mixture model for image super-resolution with multiple-focal- length degradation,” Signal Processing: Image Communication, vol. 127, p. 117139, 2024

    2024

  40. [49]

    Local light field fusion: Practical view synthesis with prescriptive sampling guidelines,

    B. Mildenhall, P. P. Srinivasan, R. Ortiz-Cayon, N. K. Kalantari, R. Ra- mamoorthi, R. Ng, and A. Kar, “Local light field fusion: Practical view synthesis with prescriptive sampling guidelines,” ACM TOG , vol. 38, no. 4, pp. 1–14, 2019

    2019

  41. [50]

    The unreasonable effectiveness of deep features as a perceptual metric,

    R. Zhang, P. Isola, A. A. Efros, E. Shechtman, and O. Wang, “The unreasonable effectiveness of deep features as a perceptual metric,” in CVPR, pp. 586–595, 2018

    2018

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.