Pith. sign in

REVIEW 5 major objections 6 minor 1 cited by

sshELF: Single-Shot Hierarchical Extrapolation of Latent Features for 3D Reconstruction from Sparse-Views

T0 review · 5 major / 6 minor · reviewed 2026-08-08 · deepseek-v4-flash

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

Pith's one-line read sshELF reconstructs a full 360-degree driving scene from six sparse outward-facing cameras in one feed-forward pass, using intermediate virtual views to fill in occluded regions instead of leaving them blurry.

desk verdict Solid two-stage sparse-view reconstruction with a genuinely useful virtual-view idea; synthetic results credible, but the nuScenes evaluation is temporally confounded and the paper overclaims on LPIPS. read the letter →

arxiv 2502.04318 v1 pith:VMNUZQOD submitted 2025-02-06 cs.CV

Eyvaz Najafli , Marius Kästingschäfer , Sebastian Bernhard , Thomas Brox , Andreas Geiger This is my paper · ORCID
classification cs.CV
keywords 3Dreconstructionsparse-viewnovelviewsynthesisGaussiansplattingvirtualviewsautonomousdrivingsceneslatentfeatureextrapolationsingle-shotfeed-forwardDINOv2features
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

sshELF is a single-shot pipeline that reconstructs a full $360^\circ$ outdoor driving scene from six outward-facing cameras with almost no overlap, in one forward pass. The paper's central claim is that the blur seen in previous sparse-view methods comes from asking the 3D decoder to do two jobs at once: extrapolate unseen content and decode it into geometry. sshELF separates these jobs by first generating intermediate virtual views through a learned cross-scene prior, then decoding reference and virtual views into 3D Gaussian primitives that can be rendered at real-time rates. On synthetic SEED4D data it reports the best PSNR among compared feed-forward baselines, and on real nuScenes data it reports higher PSNR and SSIM than MVSplat and pixelSplat at every tested temporal offset. The authors argue this shows occluded and ego-occluded regions can be reconstructed faithfully rather than left blurry, while the full pipeline runs in 0.182 seconds.

What carries the argument

The central machinery is the ELF block, a hierarchical extrapolation unit that refines virtual-view latent features by alternating epipolar-constrained cross-attention with reference-view features, self-attention, and an MLP. It is stacked four times in the backbone and once in the translator. Around it sit three load-bearing components: a pre-trained DINOv2 encoder that supplies multi-stage patch latents and a class token for both texture and semantics; a pre-trained dense depth head that initializes virtual views by projecting reference pixels into 3D and back-projecting them; and a UNet translator that decodes reference and virtual views into per-pixel 3D Gaussian primitives with probabilistic depth prediction, later rendered by Gaussian rasterization. The two stages are trained separately, with the backbone minimizing an MSE loss on reconstructed virtual-view features plus a cycle-consistency term on reference features, and the translator minimizing an MSE/MAE loss on rendered novel views and Z-buffer depth.

What would settle it

A decisive check would be to hold out a real static driving scene, hide one sensor-occluded region (such as the ground under the vehicle) from all six inputs, and compare the rendered reconstruction of that region against a dense LiDAR or drone scan; if the recovered geometry is no better than a blurry filled average, the claim that occluded regions are faithfully reconstructed would be refuted. A second check is to insert a fast-moving vehicle into an otherwise static scene and see whether the TD2–TD4 renderings degrade with the vehicle's speed, as the paper's own limitation about dynamic objects predicts.

Watch

Extended reading notes

Core claim

The paper's core discovery is that information extrapolation should be separated from primitive decoding in sparse-view 3D reconstruction: the network first hallucinates image-and-latent content for intermediate viewpoints, and only then predicts explicit 3D Gaussians from those views. This virtual-view bottleneck gives the model a place to deposit cross-scene structural priors, so the translator never has to simultaneously inpaint and reconstruct. The backbone takes six reference images, extracts multi-layer DINOv2 latent features, initializes six virtual views by depth-based point-cloud projection, and refines them with hierarchical ELF blocks built from epipolar cross-attention and self-attention, with a cycle-consistency loss that keeps reference features stable. The translator, a UNet with a single ELF block, converts the aggregated reference and virtual views into 3D Gaussian splats that are rasterized into novel views. With this design the paper reports 18.93 dB PSNR on SEED4D in 0.182 seconds and higher PSNR and SSIM than MVSplat and pixelSplat on nuScenes at all tested temporal offsets, and concludes that occluded regions are reconstructed faithfully and that jointly predicted latent features can support downstream scene understanding.

Load-bearing premise

The load-bearing premise is that ground-truth virtual views are available to supervise the backbone during training, either from synthetic exocentric cameras or by treating later timesteps of a nearly static scene as virtual views; on real nuScenes data this reduces to assuming the scene is mostly static across the temporal window, and the paper itself concedes that dynamic objects introduce transient artifacts when multi-timestep data are aggregated.

Editorial extensions

If this is right

  • A six-camera surround rig can produce a complete $360^\circ$ scene model in a single forward pass, removing the per-scene optimization step that makes iterative methods too slow for online driving use.
  • Occluded and ego-occluded regions are filled by the virtual-view stage instead of being left blurry, because representational capacity is spent on extrapolation before the 3D decoder runs.
  • Because backbone and translator train separately, raising the number or resolution of virtual views does not force a full end-to-end retrain, keeping training practical on the reported hardware.
  • The same inference pass emits latent features, texture, and depth together, enabling scene-understanding tasks to reuse the reconstruction rather than recompute features from rendered images.
  • 3D Gaussian splatting output keeps rendering real-time, so the reconstructed scene can be navigated at interactive frame rates, unlike slower triplane or volumetric single-shot baselines.

Reading between the lines

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

  • Beyond the paper, the virtual-view supervision requirement could be relaxed by generating pseudo-supervision with any strong novel-view model, which would allow training on real data without exocentric cameras.
  • Beyond the paper, because the backbone and translator are separate, the learned extrapolation prior could be paired with a different scene decoder (NeRF, triplane, or occupancy grid) without retraining the whole system from scratch.
  • Beyond the paper, the claimed faithfulness of occlusion reconstruction could be turned into a measurable claim by reporting per-region PSNR or Chamfer distance specifically inside ego-occluded and object-occluded areas, which the paper does not do.
  • Beyond the paper, the jointly predicted latent features are an untested asset: a natural next experiment is zero-shot semantic segmentation or 3D object detection from the latent codes, which would make the reconstruction useful beyond rendering.
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

5 major / 6 minor

Summary. The paper introduces sshELF, a single-shot feed-forward pipeline for reconstructing unbounded 360-degree outdoor scenes from six outward-facing input views with minimal overlap. The architecture is two-stage: a backbone encodes reference views with DINOv2, initializes intermediate 'virtual views' by depth-based projection, and refines them with hierarchical cross-/self-attention ELF blocks under an MSE loss on virtual-view latents and texture; a translator UNet then lifts the reference and virtual views into 3D Gaussian primitives that are rendered with the 3DGS rasterizer. Training assumes virtual views are available for supervision, sampled from exocentric views on SEED4D and from temporally offset egocentric views on nuScenes (TD=2,3,4). The authors report the best PSNR on SEED4D, a second rank in SSIM and D-RMSE, an 182 ms end-to-end runtime, and claim competitive or superior results on nuScenes.

Significance. If the claims hold, the paper addresses a practically important problem: real-time reconstruction of driving scenes from sparse surround cameras with almost no overlap. The virtual-view intermediate representation is a sensible mechanism for injecting global context, and decoupling extrapolation from primitive decoding is a clean design that also lowers training cost. The evaluation is standard supervised training with held-out testing, and the integration of pretrained DINOv2 features and depth priors is a strength. However, the real-world validation rests on a temporal-aggregation protocol that assumes near-static scenes, and several textual claims exceed what the reported tables support; both need correction before the contribution is fully established. The method is nevertheless a reasonable candidate for the community if those issues are addressed.

major comments (5)
  1. [Section 1 (third key insight)] The sentence 'Our results outperform previous methods' is contradicted by the paper's own Table 1: sshELF's SSIM (0.65) and LPIPS (0.50) are worse than 6Img-to-3D's (0.73 and 0.45), and pixelSplat's LPIPS (0.44) is also better. The abstract and Section 4.3 use the accurate word 'competitive' (the latter correctly notes the second rank in SSIM and D-RMSE); the introduction should be reworded to match its own data rather than claiming overall superiority.
  2. [Section 4.3 (NuScenes paragraph)] The claim that Table 2 'demonstrate[s] our method's superiority across visual and geometric metrics' is not supported by the table: sshELF's LPIPS (0.634 / 0.650 / 0.653 at TD=2/3/4) is worse than pixelSplat (0.495 / 0.532 / 0.539) at every temporal difference and worse than MVSplat at TD=3 (0.650 vs 0.575). The superiority statement holds only for PSNR, SSIM, and Chamfer; the claim should either be restricted to those metrics or the LPIPS deficit should be discussed quantitatively.
  3. [Section 4.1 (NuScenes), Section 3.5, Section 5] The nuScenes protocol aggregates egocentric views across time (TD=2,3,4) to define both virtual and novel view supervision. This treats the scene as predominantly static: moving objects appear in the target views without corresponding geometry in the reference views, so the reported errors conflate true novel-view extrapolation with failure to predict object motion. Section 5 concedes 'sensitivity to dynamic objects when aggregating multi-timestep data, which can introduce transient artifacts,' and Section 3.5 states the supervision assumption explicitly. Because this protocol is the only real-world validation, the nuScenes numbers do not yet substantiate the headline claim of faithful reconstruction of occluded regions in dynamic outdoor scenes. I recommend reporting metrics on static-region masks (or with dynamic objects masked), and re-framing the real-world claim accordingly.
  4. [Table 2 (Chamfer column)] The pixelSplat Chamfer values (1.191M, 0.144M, 0.163M at TD=2/3/4) are three to six orders of magnitude larger than every other value in the same table (e.g., 646.11 for MVSplat and 51.67 for sshELF at TD=2), which is implausible for scene-scale point cloud distances and suggests a units or normalization error. Since the text appeals to geometric metrics for superiority, these numbers must be corrected or recomputed before the comparison can be interpreted.
  5. [Section 4.2–4.3 (baselines on nuScenes)] The paper does not state how MVSplat and pixelSplat were trained or fine-tuned for the nuScenes evaluation, while sshELF is explicitly fine-tuned on nuScenes from synthetic checkpoints for 100K steps. If the baselines were evaluated with public checkpoints without nuScenes training, the Table 2 comparison is unfair; if they were trained under the same temporal-aggregation protocol, that protocol should be described. This information is needed to interpret the claimed superiority on real-world data.
minor comments (6)
  1. [Abstract and Section 1] Typos: 'hierarchal' should be 'hierarchical'; 'Our key insights is' should be 'Our key insight is'; the Section 1 heading 'Problem Statment' should be 'Problem Statement'.
  2. [Table 1] The Time column mixes seconds and milliseconds without a consistent unit label (0.42ms, 32ms, 1.1ms, 182ms, 2.85s). Also, the 182ms figure is for SEED4D at 256×256 novel view resolution, while the introduction's 0.18s claim refers to six surround views; it should be stated whether this runtime includes DINOv2 feature extraction and whether it holds at the 896×896 nuScenes rendering resolution.
  3. [Section 4.3 (NuScenes paragraph)] The phrase 'as seen in Figure 4 for MVSplat' appears to cite the wrong figure: nuScenes qualitative results are shown in Figure 5, while Figure 4 shows the SEED4D comparison.
  4. [Section 4.2] The rationale for setting λ4 = 0 on nuScenes is unclear given that LiDAR-based Chamfer supervision is used in the same section (Table 5); a brief explanation of why the Z-buffer depth loss is disabled there would help.
  5. [Sections 3.2–3.3] Several implementation details are missing: the number n of DINOv2 blocks used for latent features, the channel/resolution configuration of the ELF blocks, and the specifics of the 'probabilistic depth map prediction similar to pixelSplat' are referenced but not described.
  6. [Section 4.3 and Conclusion] The claim that sshELF 'faithfully reconstructs occluded regions' is supported only by qualitative figures; a quantitative evaluation restricted to regions unobserved in all reference views would substantiate this central claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: virtual-view supervision is ground-truth based, the novel-view targets are held-out or future-time data, and the self-citations are dataset/baseline references rather than load-bearing derivations.

full rationale

None of the seven circularity patterns applies to the paper's derivation chain. The backbone is trained with an MSE loss between reconstructed and ground-truth virtual-view latents and textures (Eq. 10), where ground truth comes from exocentric SEED4D views or from temporally aggregated nuScenes views; these supervision targets are not constructed from the model's own predicted Gaussians or from the reported evaluation metrics. The evaluation is standard supervised training with held-out Town 2 scenes and nuScenes future-time views, so the central novel-view-synthesis claim does not reduce to a fitted parameter renamed as a prediction. The explicit assumption in Section 3.5 that virtual views are available for supervision is disclosed as a data-availability assumption, not smuggled in as a conclusion. On nuScenes this assumption is implemented by temporal aggregation, and the conclusion's statement 'A current limitation is sensitivity to dynamic objects when aggregating multi-timestep data, which can introduce transient artifacts' flags the static-scene validity risk rather than hiding it; this is a domain-gap limitation, not circular reasoning. The self-citations, SEED4D as a dataset and 6Img-to-3D as a baseline, are not used to justify the architectural claim, and the method is benchmarked against external data and external components (nuScenes, DINOv2, 3DGS, epipolar-attention ideas from latentSplat). No uniqueness theorem or prior-work authority is invoked to forbid alternatives. The main correctness concern is whether temporal aggregation on dynamic real-world scenes validates the 360-degree single-shot claim, but that concern is about external validity and static-scene assumptions, not about equivalence of inputs and outputs by construction.

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

The central claim depends primarily on the availability of supervised virtual views and on the reliability of pre-trained depth and feature extractors. The loss weights and view counts are hand-set hyperparameters. No new physical entities are posited; 'virtual views' are a methodological construct.

free parameters (4)
  • Backbone loss weights lambda1, lambda2 = lambda1=1000.0, lambda2=0.1
    Chosen by hand to balance backbone feature reconstruction and cycle consistency; directly affect training.
  • Translator loss weights lambda3, lambda4 = lambda3=100.0, lambda4=0.001
    Chosen by hand to balance RGB and depth losses; on nuScenes lambda4 is set to zero.
  • Number of virtual views n_vrt = 6
    Set during cross-scene training; more or fewer would change information available to the translator.
  • Virtual view resolution = 64x64
    The backbone generates virtual views at 64x64; higher resolution improves translator input but increases cost.
assumptions (4)
  • domain assumption Ground-truth virtual views are available for supervision during training (sampled from exocentric data in SEED4D, or interpolated from temporal neighbors in nuScenes)
    Section 3.5: 'We assume the virtual views to be available for supervision, which can be sampled from the existing data in practice.' This is a key premise: the method relies on the existence of such supervisory views.
  • domain assumption DINOv2 pre-trained features retain sufficient semantic and geometric information to support view extrapolation
    Section 3.1: the image encoder is DINOv2, chosen because it 'is semantically rich and retains geometric information well'. If this fails, the whole backbone loses its information source.
  • domain assumption A pre-trained depth head (DPT) provides approximately metric depth maps that can reliably unproject reference pixels into 3D
    Section 3.2 uses 'a fine-tuned dense depth prediction transformer' to initialize virtual views; scale errors would corrupt initialization.
  • domain assumption Scene is predominantly static when novel views are taken from future timesteps (nuScenes evaluation)
    Section 4.1 builds nuScenes novel views from temporal differences; the conclusion admits sensitivity to dynamic objects, so the static-scene assumption is violated in practice.
invented entities (1)
  • Virtual views
    purpose: Intermediate viewpoints generated by the backbone to extrapolate information to unobserved regions before 3D decoding.
    Virtual views are a new intermediate representation introduced by this paper. They are not directly observable in the input nor independently verifiable; they are a latent design choice trained with supervision. Independent evidence: none outside the method.

how reviews work

0 comments
Cite this review

Pith. "Pith review of sshELF: Single-Shot Hierarchical Extrapolation of Latent Features for 3D Reconstruction from Sparse-Views." pith.science (2026). https://pith.science/paper/VMNUZQOD

@misc{pith2026250204318,
  author       = {Pith},
  title        = {Pith review of: sshELF: Single-Shot Hierarchical Extrapolation of Latent Features for 3D Reconstruction from Sparse-Views},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VMNUZQOD}},
  note         = {Machine review of arXiv:2502.04318}
}
read the original abstract

Reconstructing unbounded outdoor scenes from sparse outward-facing views poses significant challenges due to minimal view overlap. Previous methods often lack cross-scene understanding and their primitive-centric formulations overload local features to compensate for missing global context, resulting in blurriness in unseen parts of the scene. We propose sshELF, a fast, single-shot pipeline for sparse-view 3D scene reconstruction via hierarchal extrapolation of latent features. Our key insights is that disentangling information extrapolation from primitive decoding allows efficient transfer of structural patterns across training scenes. Our method: (1) learns cross-scene priors to generate intermediate virtual views to extrapolate to unobserved regions, (2) offers a two-stage network design separating virtual view generation from 3D primitive decoding for efficient training and modular model design, and (3) integrates a pre-trained foundation model for joint inference of latent features and texture, improving scene understanding and generalization. sshELF can reconstruct 360 degree scenes from six sparse input views and achieves competitive results on synthetic and real-world datasets. We find that sshELF faithfully reconstructs occluded regions, supports real-time rendering, and provides rich latent features for downstream applications. The code will be released.

Figures

Figures reproduced from arXiv: 2502.04318 by the authors.

Figure 1
Figure 1. Overview. Given a number of input images, sshELF first reconstructs several virtual views and only then predicts the 3D Gaussian primitives of the scene from which novel views are rendered. The colors of the latent information correspond to different object classes, such as purple for buildings and green for vegetation. object occlusions (areas hidden behind terrain or other ve￾hicles) and ego-occlusions (regions ob… view at source ↗
Figure 2
Figure 2. Reference, Virtual and Novel Views. An example show￾ing input views in green, a set of virtual views in red, and potential novel views in blue. Virtual view generation is key to enhancing representational capacity and extrapolating to unobserved scene areas. changes (Chen et al., 2025; Yu et al., 2021), some lack support for multi-view aggregation necessary for 360-degree surround-view synthesis, (Szymanowicz et al.… view at source ↗
Figure 3
Figure 3. Overview of sshELF. Given a few input images, sshELF first encodes them into latent features using a pre-trained DinoV2 (Sec.3.1). As part of the backbone, the latent features, together with a pre-trained depth head, are used to initialize the virtual views, which are refined using hierarchical ELF blocks consisting of cross- and self-attention layers (Sec. 3.2). Reference and virtual views are then fed into the tra… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Qualitative Novel View Synthesis Comparison on SEED4D Test Set. Comparison of large-baseline novel view synthesis under sparse observation conditions. Six ego-centric in￾put frames (top row) with limited overlap serve as reference views. We evaluate each method’s abili…
Figure 5
Figure 5. Figure 5: Qualitative Novel View Synthesis Comparison on nuScenes Test Set. Visualization of multi-view synthesis results using six reference views captured at t=0. We compare novel views reconstructed at temporal difference of TD=2, 3, and 4 (1s, 1.5s, and 2s, respectively). 4.…

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. Sparse-View 3D Reconstruction: Recent Advances and Open Challenges

    cs.CV 2025-07 conditional novelty 3.0 of 10

    A comprehensive survey that organizes sparse-view 3D reconstruction methods into geometry-based, NeRF, 3DGS, and diffusion-based categories, with benchmarks and open challenges.

Reference graph

Works this paper leans on

25 extracted references · 14 canonical work pages · cited by 1 Pith paper

  1. [4]

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

    Chen, Y ., Xu, H., Zheng, C., Zhuang, B., Pollefeys, M., Geiger, A., Cham, T.-J., and Cai, J. Mvsplat: Efficient 3d gaussian splatting from sparse multi-view images. In Leonardis, A., Ricci, E., Roth, S., Russakovsky, O., Sat- tler, T., and Varol, G. (eds.), Computer Vision – ECCV 2024, pp. 370–386, Cham,

    work page 2024

  2. [6]

    doi: 10.1109/CVPR52729.2023.01263

    IEEE Computer Society. doi: 10.1109/CVPR52729.2023.01263. Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., Uszkoreit, J., and Houlsby, N. An image is worth 16x16 words: Transformers for image recognition at scale. ICLR,

    work page arXiv 2023

  3. [9]

    VEGS: view extrapolation of urban scenes in 3d gaussian splat- ting using learned priors

    Hwang, S., Kim, M., Kang, T., Kang, J., and Choo, J. VEGS: view extrapolation of urban scenes in 3d gaussian splat- ting using learned priors. CoRR, abs/2407.02945,

    arXiv

  4. [10]

    Irshad, M

    doi: 10.48550/ARXIV .2407.02945. Irshad, M. Z., Zakharov, S., Liu, K., Guizilini, V ., Kollar, T., Gaidon, A., Kira, Z., and Ambrus, R. Neo 360: Neural fields for sparse view synthesis of outdoor scenes. Intern- taional Conference on Computer Vision (ICCV) ,

    arXiv doi:10.48550/arxiv.2407.02945

  5. [11]

    Accelerating 3d deep learning with pytorch3d

    9 sshELF: Single-Shot Hierarchical Extrapolation of Latent Features Johnson, J., Ravi, N., Reizenstein, J., Novotny, D., Tul- siani, S., Lassner, C., and Branson, S. Accelerating 3d deep learning with pytorch3d. In SIGGRAPH Asia 2020 Courses, SA ’20, New York, NY , USA,

    work page 2020

  6. [13]

    Autosplat: Constrained gaussian splatting for autonomous driving scene reconstruction

    Khan, M., Fazlali, H., Sharma, D., Cao, T., Bai, D., Ren, Y ., and Liu, B. Autosplat: Constrained gaussian splatting for autonomous driving scene reconstruction. arXiv preprint, arXiv:2407.02598,

    arXiv

  7. [14]

    Kingma, D. P. and Ba, J. Adam: A method for stochastic optimization. In Bengio, Y . and LeCun, Y . (eds.), 3rd International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Confer- ence Track Proceedings,

    work page 2015

  8. [16]

    URL https://arxiv.org/abs/2412.00730. Lefaudeux, B., Massa, F., Liskovich, D., Xiong, W., Caggiano, V ., Naren, S., Xu, M., Hu, J., Tintore, M., Zhang, S., Labatut, P., Haziza, D., Wehrstedt, L., Reizen- stein, J., and Sizov, G. xformers: A modular and hack- able transformer modelling library. https://github. com/facebookresearch/xformers,

    work page Pith review arXiv

Show all 25 references
  1. [17]

    Vdg: Vision-only dynamic gaussian for driving simulation

    Li, H., Li, J., Zhang, D., Wu, C., Shi, J., Zhao, C., Feng, H., Ding, E., Wang, J., and Han, J. Vdg: Vision-only dynamic gaussian for driving simulation. arXiv preprint, 2024a. Li, J., Zhang, J., Bai, X., Zheng, J., Ning, X., Zhou, J., and Gu, L. Dngaussian: Optimizing sparse-...

    arXiv

  2. [18]

    Oquab, M., Darcet, T., Moutakanni, T., V o, H. V ., Szafraniec, M., Khalidov, V ., Fernandez, P., Haziza, D., Massa, F., El-Nouby, A., Howes, R., Huang, P.-Y ., Xu, H., Sharma, V ., Li, S.-W., Galuba, W., Rabbat, M., As- sran, M., Ballas, N., Synnaeve, G., Misra, I., Jegou, H....

    arXiv

  3. [19]

    Common objects in 3d: Large-scale learning and evaluation of real-life 3d category reconstruction

    Reizenstein, J., Shapovalov, R., Henzler, P., Sbordone, L., Labatut, P., and Novotny, D. Common objects in 3d: Large-scale learning and evaluation of real-life 3d category reconstruction. In 2021 IEEE/CVF Interna- tional Conference on Computer Vision (ICCV), pp. 10881– 10891, ...

    2021

  4. [20]

    doi: 10.1109/ICCV48922.2021.01072

    IEEE Com- puter Society. doi: 10.1109/ICCV48922.2021.01072. Ronneberger, O., Fischer, P., and Brox, T. U-net: Con- volutional networks for biomedical image segmentation. CoRR, abs/1505.04597,

    2021

  5. [21]

    Flash3d: Feed-forward generalisable 3d scene reconstruction from a single image

    10 sshELF: Single-Shot Hierarchical Extrapolation of Latent Features Szymanowicz, S., Insafutdinov, E., Zheng, C., Campbell, D., Henriques, J., Rupprecht, C., and Vedaldi, A. Flash3d: Feed-forward generalisable 3d scene reconstruction from a single image. arxiv, 2024a. Szymano...

    arXiv

  6. [23]

    Gaussianobject: High-quality 3d object reconstruction from four views with gaussian splatting

    Yang, C., Li, S., Fang, J., Liang, R., Xie, L., Zhang, X., Shen, W., and Tian, Q. Gaussianobject: High-quality 3d object reconstruction from four views with gaussian splatting. ACM Transactions on Graphics, 2024a. Yang, J., Ivanovic, B., Litany, O., Weng, X., Kim, S. W., Li, B...

    2024

  7. [24]

    Grm: Large gaussian reconstruction model for efficient 3d reconstruction and generation

    Yinghao, X., Zifan, S., Wang, Y ., Hansheng, C., Ceyuan, Y ., Sida, P., Yujun, S., and Gordon, W. Grm: Large gaussian reconstruction model for efficient 3d reconstruction and generation. arXiv preprint, arXiv:2403.14621,

    arXiv

  8. [25]

    Sgd: Street view synthe- sis with gaussian splatting and diffusion prior

    Yu, Z., Wang, H., Yang, J., Wang, H., Xie, Z., Cai, Y ., Cao, J., Ji, Z., and Sun, M. Sgd: Street view synthe- sis with gaussian splatting and diffusion prior. ArXiv, abs/2403.20079,

    arXiv

  9. [26]

    A., Shechtman, E., and Wang, O

    Zhang, R., Isola, P., Efros, A. A., Shechtman, E., and Wang, O. The unreasonable effectiveness of deep features as a perceptual metric. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 586–595,

    2018

  10. [2004]

    Wewer, C., Raj, K., Ilg, E., Schiele, B., and Lenssen, J. E. {latentSplat}: {A}utoencoding Variational {G}aussians for Fast Generalizable {3D} Reconstruction. In Com- puter Vision – ECCV 2024 , Lecture Notes in Computer Science, Milano, Italy,

    2024

  11. [2019]

    2019.00939

    doi: 10.1109/ICCV . 2019.00939. Bhat, S. F., Birkl, R., Wofk, D., Wonka, P., and M¨uller, M. Zoedepth: Zero-shot transfer by combining relative and metric depth. CoRR, abs/2302.12288,

    2019

  12. [2020]

    ISBN 9781450381123

    Association for Computing Machinery. ISBN 9781450381123. doi: 10.1145/3415263.3419160. Ke, B., Obukhov, A., Huang, S., Metzger, N., Daudt, R. C., and Schindler, K. Repurposing diffusion-based image generators for monocular depth estimation. In Proceed- ings of the IEEE/CVF Con...

  13. [2021]

    Lara: Efficient large-baseline radiance fields

    Chen, A., Xu, H., Esposito, S., Tang, S., and Geiger, A. Lara: Efficient large-baseline radiance fields. In Euro- pean Conference on Computer Vision (ECCV) , 2024a. Chen, Y ., Gu, C., Jiang, J., Zhu, X., and Zhang, L. Periodic vibration gaussian: Dynamic urban scene reconstruc...

    arXiv

  14. [2022]

    Guo, J., Deng, N., Li, X., Bai, Y ., Shi, B., Wang, C., Ding, C., Wang, D., and Li, Y

    doi: 10.1109/LRA.2022.3150884. Guo, J., Deng, N., Li, X., Bai, Y ., Shi, B., Wang, C., Ding, C., Wang, D., and Li, Y . Streetsurf: Extending multi- view implicit surface reconstruction to street views.arXiv preprint arXiv:2306.04988,

    2022

  15. [2023]

    Semantickitti: A dataset for semantic scene understanding of lidar sequences

    Behley, J., Garbade, M., Milioto, A., Quenzel, J., Behnke, S., Stachniss, C., and Gall, J. Semantickitti: A dataset for semantic scene understanding of lidar sequences. In 2019 IEEE/CVF International Conference on Computer Vision (ICCV), pp. 9296–9306,

    2019

  16. [2024]

    6img-to-3d: Few-image large-scale outdoor driving scene reconstruction

    Gieruc, T., K¨astingsch¨afer, M., Bernhard, S., and Salzmann, M. 6img-to-3d: Few-image large-scale outdoor driving scene reconstruction. arXiv preprint, arXiv:2404.12378,

    arXiv

  17. [2025]

    Chen, Z., Yang, J., Huang, J., Lutio, R

    Springer Nature Switzer- land. Chen, Z., Yang, J., Huang, J., Lutio, R. d., Esturo, J. M., Ivanovic, B., Litany, O., Gojcic, Z., Fidler, S., Pavone, M., Song, L., and Wang, Y . Omnire: Omni urban scene reconstruction. arXiv preprint arXiv:2408.16760, 2024b. Deitke, M., Schwenk...

    2023 arXiv

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.