Pith. sign in

REVIEW 3 major objections 5 minor 108 references

GeoScene shows that satellite imagery plus OpenStreetMap, used as learnable soft priorities rather than hard constraints, improves camera-based 3D semantic scene completion, most on large static structures such as roads and buildings.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

GeoScene uses weighted fusion of satellite imagery and OpenStreetMap priors to improve camera-based 3D semantic scene completion on SemanticKITTI and SSCBench-KITTI-360.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection A solid, well-ablated empirical SSC paper that shows a real geospatial-prior benefit, but the headline cross-method SOTA claim rests on underspecified controlled baselines and no code; worth refereeing with a request for the fusion details and variance. the 3 major comments →

arxiv 2608.03618 v1 pith:A6LS5VVY submitted 2026-08-04 cs.CV

Geospatial-Prior Guidance for 3D Semantic Scene Completion

Meng Wang , Shougao Zhang , Wenzhe He , Ruihui Li , Nan Hu , Zhuo Tang , Kenli Li This is my paper
classification cs.CV
keywords 3D semantic scene completiongeospatial priorssatellite imageryOpenStreetMapsoft reliability weightingvoxel refinementcamera-based perceptionautonomous driving
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to establish that a camera-only 3D semantic scene completion system can be pushed past its fundamental blind spots — occluded regions and areas outside the camera frustum — by handing it two freely available geospatial inputs: a satellite image and an OpenStreetMap layout of the surrounding block. The central design decision is to treat these inputs as soft priors whose influence is learned voxel by voxel (each 3D grid cell), not as ground truth to obey. GeoScene learns one reliability weight for onboard visual evidence and a second, complementary weight for geospatial guidance, then uses the pair to fuse the two feature streams so that structure the camera cannot see is filled in only where the map is trusted. On both benchmarks the framework reports the best numbers under the geospatial-prior-assisted setting, with gains concentrated on exactly the classes camera-only methods struggle with most: roads, sidewalks, buildings, vegetation and terrain. If the claim holds, practical scene completion no longer needs to invent large-scale structure from scratch — widely available map data supplies it, as long as alignment is decent.

Core claim

GeoScene claims that two free geospatial inputs — a satellite crop and an OpenStreetMap layout of the surrounding 51.2 m block — can guide camera-based 3D semantic scene completion, if treated as soft, weighted suggestions rather than hard constraints. The framework learns two complementary per-voxel reliability fields: an observation weight marking where onboard evidence is strong, and a geospatial weight letting map structure fill occluded regions. It reports 46.58% IoU / 18.76% mIoU / 32.56% Geo-mIoU on SemanticKITTI and 49.46% / 21.59% / 34.89% on SSCBench-KITTI-360, with building IoU up ≈4 points on both benchmarks. It tolerates imperfect priors: 0.5 m or 1 degree alignment error costs

What carries the argument

The carrying mechanism is the Dual-Priors Weighted Classifier (DPWC), which emits two sigmoid soft-weight volumes — W_obs marking where onboard evidence is strong, W_geo gating map structure into weakly observed regions — followed by a complementary-alignment step that normalizes the pair. The Weights-Guided Voxel Refiner (WGVR) then runs an observation branch (small convolutions, street detail) and a geospatial branch (larger convolutions, structural context), fuses both under the learned weights plus a normalized blend, and harmonizes globally. A third component, the Geospatial BEV Projection, supervises a bird's-eye-view head from a priority-ordered projection of the 3D labels. Nothing is

Load-bearing premise

The whole gain rests on the satellite and OpenStreetMap crop matching the driving scene: both are fetched once (27 January 2026) and aligned to each frame by the benchmark's GPS/IMU pose, and the paper's own perturbation tables show the benefit erodes once alignment error passes roughly 1 m or 2 degrees — if the geospatial data are missing, outdated, or misaligned, the method falls back toward camera-only quality.

What would settle it

Take the trained model and feed it geospatial tiles from a different part of the city while preserving the GPS/IMU registration: if mIoU stays near the aligned-prior 18.76%, the model is exploiting the prior's presence and statistics rather than its content, and the guidance claim is falsified; if mIoU drops toward the camera-only baseline, the weights genuinely route correct map content into unobserved voxels. The paper's zero-dropout and perturbation experiments remove or misalign priors but never substitute confidently wrong ones, so this swap test is the clean separator.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Largest, consistent gains land on the classes camera-only SSC handles worst: roads, sidewalks, buildings, vegetation and terrain improve on both benchmarks, with building IoU up about 4 points — map priors supply exactly the large-scale structure that occlusion hides.
  • The benefit survives realistic localization noise: 0.5 m or 1° GPS/IMU error costs only ~0.5 mIoU, so a vehicle with ordinary pose estimates still gains from the priors.
  • Intermittent loss of map coverage is non-fatal: 10–40% inference-time dropout of satellite/OSM inputs costs 0.4–1.25 mIoU without retraining, and 5 m / 10° alignment errors leave the onboard branch as a working fallback.
  • Every component earns its gain in ablation: satellite +0.34 mIoU, OSM +1.14, dual weight fields +0.89, voxel refiner +0.89 — a total of +2.78 mIoU and +2.45 IoU over the camera-only base.
  • The added accuracy does not come at a latency price: 0.186 s per scene at 82 M parameters, below the geospatial-augmented baselines it beats.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The paper never tests confidently wrong priors. Swapping in satellite/OSM tiles from another neighbourhood — same crop size, same alignment — would separate 'the model reads map content' from 'the model exploits the prior's presence'. The zero-dropout test removes priors but never contradicts them; a wrong-map test is the clean separator.
  • Deployment reading: the headline numbers are for the geospatial-prior-assisted setting, where every baseline is retrained with fused priors. Without usable map coverage the honest expectation is the base model's ~44 IoU / 16 mIoU — the priors are a conditional upgrade, not a free lunch.
  • W_geo is itself a usable product: voxels with high geospatial weight and near-zero observation weight are exactly where the model asserts structure it cannot see. A motion planner could read that field as a 'map-informed, unverified' mask — a safety-relevant byproduct the paper does not discuss.
  • The limitations section's own suggestions — timestamped map snapshots and map-change detection — point to a natural extension: let W_geo learn to drop outdated structures (a building demolished between map fetch and drive), turning the current robustness gap into a training signal.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes GeoScene, a camera-based 3D semantic scene completion (SSC) framework that fuses onboard RGB features with satellite imagery and OpenStreetMap (OSM) semantic priors. The core idea is to learn separate voxel-wise soft reliability weights for onboard observations and geospatial guidance (Dual-Priors Weighted Classifier) and to use these weights in a Weights-Guided Voxel Refiner to adaptively combine and refine the two feature streams. A BEV auxiliary loss, built via a hand-designed geospatial-priority projection of the 3D labels, provides additional supervision. Experiments on SemanticKITTI and SSCBench-KITTI-360 report state-of-the-art results under a 'geospatial-prior-assisted' setting, together with ablations of the components, robustness to synthetic GPS/IMU errors, and inference-time geospatial dropout.

Significance. If the results hold, GeoScene is a useful contribution to camera-based SSC: it is one of the first methods to combine satellite imagery with structured OSM cues as soft priors, and it demonstrates that learned reliability weighting can prevent hard-constraint failures when external priors are noisy or incomplete. The paper has clear strengths: the method is well specified, the ablation study is thorough (architecture components, OSM categories, BEV projection strategy, fusion operators, loss coefficients), the robustness analysis with translation/heading perturbations and geospatial dropout is valuable, and the supplementary material provides detailed data provenance and implementation settings. The main weakness is that the controlled comparison against geospatially augmented baselines, which carries the claim of a 'stronger ability to exploit geospatial priors,' is underspecified and therefore not independently checkable as written. The internal ablation (Table 4) does support the existence of a geospatial-prior benefit, so this concern is about the cross-method SOTA claim rather than the core method.

major comments (3)
  1. [Table 3, 'Quantitative Results'] The controlled comparison supporting the headline 'stronger ability to exploit geospatial priors' is not reproducible as reported. The text says each starred baseline incorporates the same satellite/OSM priors through a 'lightweight feature-fusion layer' while retaining the original baseline architecture, but the layer's architecture, placement, channel dimensions, training schedule, and hyperparameters are not specified anywhere in the paper or appendix, and no code or weights are released. Without this information, the 2.95-point Geo-mIoU gap over CGFormer* (and the 2.43-point gap over VLScene*) could reflect a weak fusion baseline rather than an intrinsic advantage of DPWC+WGVR. I am not disputing the internal ablation in Table 4, which shows a real gain over the camera-only base; the issue is specifically the cross-method SOTA claim. Please specify the fusion layer in full, report it
  2. [Tables 1-2, 'Quantitative Results'] All reported results are from a single training run, and several headline improvements are small: +0.18 IoU and +0.93 mIoU on SemanticKITTI, and +1.27 IoU on SSCBench-KITTI-360. For deep SSC models, run-to-run variation from initialization and data ordering can be of this magnitude. The absence of error bars or multiple-seed statistics makes it impossible to judge whether the differences are significant. Please report at least three seeds with mean and standard deviation for the main comparisons (Tables 1-3), or provide a statistical test. This is needed to support the word 'consistently' in the abstract and conclusion.
  3. [Eq. (16), 'Geospatial data and alignment'] The whole approach assumes the benchmark GPS/IMU pose accurately registers satellite and OSM data to the onboard frame. The appendix discloses that Mapbox does not expose acquisition timestamps and OSM was retrieved at a single date, so real temporal mismatch is not modeled. Tables 6-7 show that Geo-mIoU drops from 32.56 to 30.71 at 1 m translation error and to 30.35 at 2 degrees rotation; these are plausible localization errors in urban canyons. This does not invalidate the method, but it should be stated in the main text that the reported gains are conditional on alignment accuracy better than roughly 1 m / 2 degrees, and the real-data temporal-mismatch case (e.g., a sequence with known construction or map change) should be evaluated if available.
minor comments (5)
  1. [Figure 2] The framework diagram contains literal '????' placeholders after the BEV Header and SSC Header. These should be replaced with the actual module names or removed.
  2. [Eqs. (1)-(5)] The query proposal P is used as both a binary mask (Eq. 5) and an input to a 3D CNN (Eq. 1) and AvgConv3D (Eq. 2). Please state explicitly whether P is a binary voxel indicator or a continuous feature volume, and define its construction precisely.
  3. [Table 3] SGFormer is listed without a star, unlike the other augmented methods. Since SGFormer is itself a satellite-ground fusion method, clarify why it is not augmented with the same OSM priors and whether its numbers are taken from Table 1 or retrained under the same protocol.
  4. [Appendix, 'Geo-mIoU class definition'] The Geo-mIoU metric is defined only in the appendix, yet it is a headline metric in Tables 1-3. Move a one-sentence definition of the class subset to the main text, or at least reference the appendix at first mention.
  5. [Training Loss, Eq. (14)] The sensitivity analysis covers lambda_bev thoroughly (Table 9), but the other loss weights (lambda_s, lambda_g, lambda_ce, lambda_d) are set 'following preliminary experiments' without any supporting ablation. Consider reporting a small sensitivity study or a sentence justifying the default values.

Circularity Check

0 steps flagged

No significant circularity: empirical deep-learning paper with external geospatial inputs, public-benchmark evaluation, and no derivation reducing to its own inputs; minor self-citations are not load-bearing.

full rationale

GeoScene is an architecture-and-training paper, not a derivation. Its claimed result—improved 3D SSC with satellite/OSM priors—is supported by supervised losses (Eqs. 13–15) and external benchmark comparisons (Tables 1–2), not by a chain of equations that assumes the conclusion. The geospatial inputs (Mapbox satellite crops, OSM vectors) are external and aligned via GPS/IMU (Eq. 16); they are not constructed from the model's own outputs. The BEV auxiliary target in Eq. 13 is projected from ground-truth 3D labels; this is standard supervised auxiliary supervision, not a fitted parameter renamed as a prediction. Table 3's augmented baselines are a controlled comparison; the underspecified 'lightweight feature-fusion layer' is a reproducibility/verifiability weakness (correctness risk), not circularity, because the comparison does not reduce GeoScene's score to its own definition. Self-citations exist (e.g., 'processed by a Sparse Encoder (Wang et al. 2025b)' and VLScene baselines), but they are used as components or prior methods, not as an unverified premise that forces the result. The paper's own robustness tables (6–7) and limitation paragraph acknowledge sensitivity to alignment and OSM incompleteness, which is consistent with treating priors as external inputs. Therefore no circular step is exhibited; score reflects only minor non-load-bearing self-citation.

Axiom & Free-Parameter Ledger

8 free parameters · 4 axioms · 0 invented entities

The central claim relies on standard supervised deep network training plus the tuned hyperparameters listed above, and on the domain assumption that publicly retrieved geospatial data are aligned and current. The BEV projection priority is a manually defined heuristic. No new physical or mathematical entities are introduced.

free parameters (8)
  • lambda_s (Eq. 14) = 1
    Weight on the scaled semantic loss term; chosen via preliminary experiments.
  • lambda_g (Eq. 14) = 1
    Weight on the scaled geometric loss term; chosen via preliminary experiments.
  • lambda_ce (Eq. 14) = 3
    Weight on cross-entropy loss; tuned on validation.
  • lambda_d (Eq. 14) = 1
    Weight on depth loss; chosen via preliminary experiments.
  • lambda_bev (Eq. 15) = 1.5
    Weight on BEV auxiliary loss; sensitivity shown in Table 9.
  • r (Eq. 4) = 1.5
    Scaling coefficient for road/building regions; tuned on validation, Table 10.
  • o (Eq. 4) = 1.0
    Scaling coefficient for other regions; tuned on validation, Table 10.
  • alpha (Eq. 12) = 0.7
    Fusion coefficient between weighted and normalized fusion; tuned, Table 11.
axioms (4)
  • standard math LSS depth distribution lifting and deformable cross-attention (Philion and Fidler 2020; Zhu et al. 2020) provide a valid view transformation.
    Used as backbone components without rederivation; they are widely accepted in the SSC literature.
  • domain assumption GPS/IMU pose from the benchmarks correctly aligns satellite and OSM data to the ego frame via Eq. (16).
    Load-bearing for all geospatial inputs; robustness is tested in Tables 6 and 7 but not guaranteed for new environments.
  • domain assumption OSM primitives map to only three coarse categories (road, building, others) and are treated as time-invariant static structure.
    Dynamic objects are explicitly excluded from the geospatial branch; if OSM is outdated, the guidance is wrong, as admitted in the limitations.
  • ad hoc to paper The hand-defined priority ordering in BEV projection (tall structures, mid-level elements, ground-level classes) yields useful training targets.
    No theoretical justification; empirically chosen and supported only by the ablation in Table 14.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Geospatial-Prior Guidance for 3D Semantic Scene Completion." pith.science (2026). https://pith.science/paper/A6LS5VVY

@misc{pith2026260803618,
  author       = {Pith},
  title        = {Pith review of: Geospatial-Prior Guidance for 3D Semantic Scene Completion},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/A6LS5VVY}},
  note         = {Machine review of arXiv:2608.03618}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Inferring complete 3D geometry and semantics from onboard images remains challenging because occlusions and restricted fields of view leave large scene regions underconstrained. Although satellite imagery provides wide-area context, appearance cues alone offer limited structural guidance and may be unreliable because of spatial or temporal discrepancies. We present GeoScene, a geospatially guided framework that jointly uses satellite imagery and structured OpenStreetMap cues as soft priors for 3D semantic scene completion. GeoScene learns complementary voxel-wise reliability weights for onboard observations and geospatial guidance, and uses them to control feature refinement in observed and unobserved regions. This design preserves local visual evidence while exploiting large-scale road and building structure beyond onboard visibility. Experiments on SemanticKITTI and SSCBench-KITTI-360 demonstrate that GeoScene consistently improves both geometric and semantic completion under the geospatial-prior-assisted setting, with the most pronounced benefits for large-scale static and geospatially structured classes.

Figures

Figures reproduced from arXiv: 2608.03618 by Kenli Li, Meng Wang, Nan Hu, Ruihui Li, Shougao Zhang, Wenzhe He, Zhuo Tang.

Figure 1
Figure 1. Figure 1: GeoScene integrates geospatial priors to guide 3D semantic scene completion. Compared with previous SSC methods, [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The GeoScene framework for geospatially guided 3D semantic scene completion. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Qualitative comparisons on the SemanticKITTI validation set. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Detailed architecture of GeoScene. (a) Geo-Aware View Transformation projects observation and geospatial features [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Qualitative comparison of BEV target projection strategies. The geospatial-priority strategy retains stable static [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗
Figure 7
Figure 7. Figure 7: Failure case under missing OSM coverage. High [PITH_FULL_IMAGE:figures/full_fig_p013_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Additional qualitative comparisons on the SemanticKITTI validation set. GeoScene produces more continuous large [PITH_FULL_IMAGE:figures/full_fig_p014_8.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

108 extracted references · 54 canonical work pages · 2 internal anchors

  1. [1]

    Communication, Simulation, and Intelligent Agents: Implications of Personal Intelligent Machines for Medical Education

    Clancey, William J. Communication, Simulation, and Intelligent Agents: Implications of Personal Intelligent Machines for Medical Education. Proceedings of the Eighth International Joint Conference on Artificial Intelligence (IJCAI-83)

  2. [2]

    Classification Problem Solving

    Clancey, William J. Classification Problem Solving. Proceedings of the Fourth National Conference on Artificial Intelligence

  3. [3]

    , title =

    Robinson, Arthur L. , title =. 1980 , doi =. https://science.sciencemag.org/content/208/4447/1019.full.pdf , journal =

    1980

  4. [4]

    New Ways to Make Microcircuits Smaller---Duplicate Entry

    Robinson, Arthur L. New Ways to Make Microcircuits Smaller---Duplicate Entry. Science

  5. [5]

    Clancey and Glenn Rennels , abstract =

    Diane Warner Hasling and William J. Clancey and Glenn Rennels , abstract =. Strategic explanations for a diagnostic consultation system , journal =. 1984 , issn =. doi:https://doi.org/10.1016/S0020-7373(84)80003-6 , url =

    doi:10.1016/s0020-7373(84)80003-6 1984

  6. [6]

    and Rennels, Glenn R

    Hasling, Diane Warner and Clancey, William J. and Rennels, Glenn R. and Test, Thomas. Strategic Explanations in Consultation---Duplicate. The International Journal of Man-Machine Studies

  7. [7]

    Poligon: A System for Parallel Problem Solving

    Rice, James. Poligon: A System for Parallel Problem Solving

  8. [8]

    Transfer of Rule-Based Expertise through a Tutorial Dialogue

    Clancey, William J. Transfer of Rule-Based Expertise through a Tutorial Dialogue

  9. [9]

    The Engineering of Qualitative Models

    Clancey, William J. The Engineering of Qualitative Models

  10. [10]

    2023 , eprint=

    Attention Is All You Need , author=. 2023 , eprint=

    2023

  11. [11]

    Pluto: The 'Other' Red Planet

    NASA. Pluto: The 'Other' Red Planet

  12. [12]

    CVPR , year=

    MonoScene: Monocular 3D Semantic Scene Completion , author=. CVPR , year=

  13. [13]

    CVPR , pages=

    Tri-perspective view for vision-based 3d semantic occupancy prediction , author=. CVPR , pages=

  14. [14]

    CVPR , year=

    VoxFormer: Sparse Voxel Transformer for Camera-based 3D Semantic Scene Completion , author=. CVPR , year=

  15. [15]

    arXiv preprint arXiv:2304.05316 , year=

    OccFormer: Dual-path Transformer for Vision-based 3D Semantic Occupancy Prediction , author=. arXiv preprint arXiv:2304.05316 , year=

    Pith/arXiv arXiv

  16. [16]

    CVPR , pages=

    Semantic scene completion from a single depth image , author=. CVPR , pages=

  17. [17]

    TPAMI , volume=

    Semantic scene completion using local deep implicit functions on lidar data , author=. TPAMI , volume=. 2021 , publisher=

    2021

  18. [18]

    ECCV , pages=

    Efficient semantic scene completion network with spatial group convolution , author=. ECCV , pages=

  19. [19]

    CVPR , pages=

    Semantic scene completion via integrating instances and scene in-the-loop , author=. CVPR , pages=

  20. [20]

    CVPR , pages=

    Rgbd based dimensional decomposition residual network for 3d semantic scene completion , author=. CVPR , pages=

  21. [21]

    ECCV , pages=

    Lift, splat, shoot: Encoding images from arbitrary camera rigs by implicitly unprojecting to 3d , author=. ECCV , pages=. 2020 , organization=

    2020

  22. [22]

    International journal of computer vision , volume=

    Imagenet large scale visual recognition challenge , author=. International journal of computer vision , volume=. 2015 , publisher=

    2015

  23. [23]

    ECCV , pages=

    Bevformer: Learning bird’s-eye-view representation from multi-camera images via spatiotemporal transformers , author=. ECCV , pages=. 2022 , organization=

    2022

  24. [24]

    arXiv preprint arXiv:1811.08188 , year=

    Orthographic feature transform for monocular 3d object detection , author=. arXiv preprint arXiv:1811.08188 , year=

    Pith/arXiv arXiv

  25. [25]

    Conference on Robot Learning , pages=

    Detr3d: 3d object detection from multi-view images via 3d-to-2d queries , author=. Conference on Robot Learning , pages=. 2022 , organization=

    2022

  26. [26]

    AAAI , volume=

    Bevdepth: Acquisition of reliable depth for multi-view 3d object detection , author=. AAAI , volume=

  27. [27]

    ICCV , pages=

    Semantickitti: A dataset for semantic scene understanding of lidar sequences , author=. ICCV , pages=

  28. [28]

    ICML , pages=

    Efficientnet: Rethinking model scaling for convolutional neural networks , author=. ICML , pages=. 2019 , organization=

    2019

  29. [29]

    ICML , pages=

    Global context vision transformers , author=. ICML , pages=. 2023 , organization=

    2023

  30. [30]

    ICLR , year=

    Deformable DETR: Deformable Transformers for End-to-End Object Detection , author=. ICLR , year=

  31. [31]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    VLScene: Vision-Language Guidance Distillation for Camera-Based 3D Semantic Scene Completion , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  32. [32]

    arXiv preprint arXiv:1711.05101 , year=

    Decoupled weight decay regularization , author=. arXiv preprint arXiv:1711.05101 , year=

    Pith/arXiv arXiv

  33. [33]

    Li, Zhiqi and Yu, Zhiding and Wang, Wenhai and Anandkumar, Anima and Lu, Tong and Alvarez, Jose M , booktitle=

    work page

  34. [34]

    IJCAI , pages=

    View-volume network for semantic scene completion from a single depth image , author=. IJCAI , pages=

    work page

  35. [35]

    ICCV , pages=

    Swin transformer: Hierarchical vision transformer using shifted windows , author=. ICCV , pages=

  36. [36]

    arXiv preprint arXiv:2010.11929 , year=

    An image is worth 16x16 words: Transformers for image recognition at scale , author=. arXiv preprint arXiv:2010.11929 , year=

    Pith/arXiv arXiv 2010

  37. [37]

    Advances in neural information processing systems , volume=

    Attention is all you need , author=. Advances in neural information processing systems , volume=

  38. [38]

    CVPR , pages=

    Anisotropic convolutional networks for 3d semantic scene completion , author=. CVPR , pages=

    work page

  39. [39]

    3DV , pages=

    Lmscnet: Lightweight multiscale 3d semantic completion , author=. 3DV , pages=. 2020 , organization=

    work page 2020

  40. [40]

    AAAI , volume=

    Sparse single sweep lidar point cloud segmentation via learning contextual shape priors from scene completion , author=. AAAI , volume=

    work page

  41. [41]

    CVPR , pages=

    Feature pyramid networks for object detection , author=. CVPR , pages=

    work page

  42. [42]

    ICCV , pages=

    Surroundocc: Multi-camera 3d occupancy prediction for autonomous driving , author=. ICCV , pages=

    work page

  43. [43]

    ICCV , pages=

    Ndc-scene: Boost monocular 3d semantic scene completion in normalized device coordinates space , author=. ICCV , pages=

    work page

  44. [44]

    arXiv preprint arXiv:2306.09001 , year=

    SSCBench: Monocular 3D Semantic Scene Completion Benchmark in Street Views , author=. arXiv preprint arXiv:2306.09001 , year=

    Pith/arXiv arXiv

  45. [45]

    CVPR , year =

    Andreas Geiger and Philip Lenz and Raquel Urtasun , title =. CVPR , year =

  46. [46]

    Yiyi Liao and Jun Xie and Andreas Geiger , journal =

    work page

  47. [47]

    2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages=

    Up-to-down network: Fusing multi-scale context for 3d semantic scene completion , author=. 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages=. 2021 , organization=

    work page 2021

  48. [48]

    2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages=

    Semantic segmentation-assisted scene completion for lidar point clouds , author=. 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages=. 2021 , organization=

    work page 2021

  49. [49]

    2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages=

    SSC-RS: Elevate LiDAR Semantic Scene Completion with Representation Separation and BEV Fusion , author=. 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages=. 2023 , organization=

    work page 2023

  50. [50]

    arXiv preprint arXiv:2303.13959 , year=

    Stereoscene: Bev-assisted stereo matching empowers 3d semantic scene completion , author=. arXiv preprint arXiv:2303.13959 , year=

    Pith/arXiv arXiv

  51. [51]

    Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

    Computing the stereo matching cost with a convolutional neural network , author=. Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

    work page

  52. [52]

    Proceedings of the European conference on computer vision (ECCV) , pages=

    Mvsnet: Depth inference for unstructured multi-view stereo , author=. Proceedings of the European conference on computer vision (ECCV) , pages=

  53. [53]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Adabins: Depth estimation using adaptive bins , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  54. [54]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , month =

    Yu, Zehao and Gao, Shenghua , title =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , month =

    work page

  55. [55]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , month =

    Bae, Gwangbin and Budvytis, Ignas and Cipolla, Roberto , title =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , month =. 2022 , pages =

    work page 2022

  56. [56]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Bevstereo: Enhancing depth estimation in multi-view 3d object detection with temporal stereo , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    work page

  57. [57]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Group-wise correlation stereo network , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  58. [58]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Cascade cost volume for high-resolution multi-view stereo and stereo matching , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  59. [59]

    IEEE Transactions on Pattern Analysis and Machine Intelligence , volume=

    Towards accurate reconstruction of 3d scene shape from a single monocular image , author=. IEEE Transactions on Pattern Analysis and Machine Intelligence , volume=. 2022 , publisher=

    work page 2022

  60. [60]

    Advances in Neural Information Processing Systems , year =

    RoboDepth: Robust Out-of-Distribution Depth Estimation under Corruptions , author =. Advances in Neural Information Processing Systems , year =

    work page

  61. [61]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , month =

    Li, Rui and Gong, Dong and Yin, Wei and Chen, Hao and Zhu, Yu and Wang, Kaixuan and Chen, Xiaozhi and Sun, Jinqiu and Zhang, Yanning , title =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , month =. 2023 , pages =

    work page 2023

  62. [62]

    IEEE transactions on pattern analysis and machine intelligence , volume=

    Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs , author=. IEEE transactions on pattern analysis and machine intelligence , volume=. 2017 , publisher=

    2017

  63. [63]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , month =

    Hassani, Ali and Walton, Steven and Li, Jiachen and Li, Shen and Shi, Humphrey , title =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , month =. 2023 , pages =

    work page 2023

  64. [64]

    International conference on machine learning , pages=

    Scaling up visual and vision-language representation learning with noisy text supervision , author=. International conference on machine learning , pages=. 2021 , organization=

    2021

  65. [65]

    International conference on machine learning , pages=

    Learning transferable visual models from natural language supervision , author=. International conference on machine learning , pages=. 2021 , organization=

    2021

  66. [66]

    arXiv preprint arXiv:2201.03546 , year=

    Language-driven semantic segmentation , author=. arXiv preprint arXiv:2201.03546 , year=

    Pith/arXiv arXiv

  67. [67]

    European Conference on Computer Vision , pages=

    Scaling open-vocabulary image segmentation with image-level labels , author=. European Conference on Computer Vision , pages=. 2022 , organization=

    work page 2022

  68. [68]

    European Conference on Computer Vision , pages=

    Bottom up top down detection transformers for language grounding in images and point clouds , author=. European Conference on Computer Vision , pages=. 2022 , organization=

    2022

  69. [69]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    Lerf: Language embedded radiance fields , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    work page

  70. [70]

    OpenScene: 3D Scene Understanding with Open Vocabularies , author =

    work page

  71. [71]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Clip2scene: Towards label-efficient 3d scene understanding by clip , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    work page

  72. [72]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Pla: Language-driven open-vocabulary 3d scene understanding , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  73. [73]

    The Eleventh International Conference on Learning Representations , year=

    Open-Vocabulary Object Detection upon Frozen Vision and Language Models , author=. The Eleventh International Conference on Learning Representations , year=

    work page

  74. [74]

    arXiv preprint arXiv:2304.00962 , year=

    Regionplc: Regional point-language contrastive learning for open-world 3d scene understanding , author=. arXiv preprint arXiv:2304.00962 , year=

    Pith/arXiv arXiv

  75. [75]

    International Journal of Computer Vision , volume=

    Clip-adapter: Better vision-language models with feature adapters , author=. International Journal of Computer Vision , volume=. 2024 , publisher=

    2024

  76. [76]

    Enhanced Geometry and Semantics for Camera-Based 3D Semantic Scene Completion , year=

    Xiao, Haihong and Kang, Wenxiong and Guo, Yulan and Liu, Hao and He, Ying , journal=. Enhanced Geometry and Semantics for Camera-Based 3D Semantic Scene Completion , year=

    work page

  77. [77]

    WSSIC-Net: Weakly-Supervised Semantic Instance Completion of 3D Point Cloud Scenes , year=

    Fu, Zhiheng and Guo, Yulan and Chen, Minglin and Hu, Qingyong and Laga, Hamid and Boussaid, Farid and Bennamoun, Mohammed , journal=. WSSIC-Net: Weakly-Supervised Semantic Instance Completion of 3D Point Cloud Scenes , year=

    work page

  78. [78]

    European Conference on Computer Vision , pages=

    2dpass: 2d priors assisted semantic segmentation on lidar point clouds , author=. European Conference on Computer Vision , pages=. 2022 , organization=

    work page 2022

  79. [79]

    IEEE Transactions on Image Processing , volume=

    Efficient 3D point cloud feature learning for large-scale place recognition , author=. IEEE Transactions on Image Processing , volume=. 2022 , publisher=

    work page 2022

  80. [80]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    OctOcc: High-Resolution 3D Occupancy Prediction with Octree , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    work page

Showing first 80 references.

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.