Pith. sign in

REVIEW 3 major objections 6 minor 50 references

Depth-grounded Gaussians improve vision-only 3D occupancy prediction, reaching 34.07 SC IoU on nuScenes.

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 ↗

T0 review · deepseek-v4-flash

2026-08-01 16:08 UTC pith:OOZ7AK5G

load-bearing objection VGOcc is a competent combination of foundation features and sparse Gaussian occupancy with clean ablations, but its SOTA claim is unproven because the closest baselines are missing from the comparison. the 3 major comments →

arxiv 2607.18078 v1 pith:OOZ7AK5G submitted 2026-07-20 cs.CV

VGOcc: Learning Visual-Geometric Gaussians for Vision-Centric 3D Driving Occupancy Prediction

Junhong Lin , Xianda Guo , Kangli Wang , Yuqi Ye , Xiaoyu Liang , Yanlun Peng , Wei Gao This is my paper
classification cs.CV
keywords 3D semantic occupancy predictionvision-centric perceptionsparse Gaussiansoccupancy predictionautonomous drivingvisual-geometric foundation modelsray-depthcross-view feature learning
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.

This paper tries to establish that sparse Gaussian primitives for camera-only 3D semantic occupancy prediction become substantially more accurate when their initialization and refinement are grounded in complementary geometric and visual-semantic cues from pretrained foundation models, rather than relying on image features alone. It introduces Visual-Geometric Gaussians, which place Gaussian centers from predicted per-ray depth hypotheses with balanced spatial sampling, initialize attributes from same-ray visual features, and refine using pose-conditioned, cross-view awareness. On the nuScenes benchmark, VGOcc reports 34.07 SC IoU and 21.75 SSC mIoU, surpassing the previous best sparse-Gaussian method by 3.51 and 1.73 points, and outperforming dense voxel and BEV methods as well. The gain matters because occupancy prediction for autonomous driving needs efficient yet geometrically faithful 3D reconstruction from ordinary cameras.

Core claim

VGOcc argues that sparse Gaussian primitives—small ellipsoids with semantic attributes—are a good representation for 3D occupancy only when their birth and refinement are steered by explicit geometry. The paper therefore builds Visual-Geometric Gaussians: ray-depth hypotheses from a frozen geometric backbone are lifted to 3D, selected by fast voxel-thinned sampling to avoid clustering on visible surfaces, and paired with same-ray visual-semantic features that initialize each primitive's opacity, class, and latent code. During refinement, a pose-aware feature module conditions multi-level tokens on each patch's calibrated ray and aggregates neighboring-view features at projected 3D points, in

What carries the argument

Visual-Geometric Gaussians — the central object. They form a set of sparse Gaussian primitives whose placement comes from geometry (ray-depth hypotheses plus voxel-thinned sampling), whose non-spatial attributes come from same-ray visual features, and whose refinement features are pose-conditioned fusion of semantic, geometric, and global tokens with cross-view ray compatibility. This machinery carries the argument because the ablations show the largest drops when the geometric birth is replaced by previous query initialization (1.65/1.51 IoU/mIoU) and when pose conditioning is removed (1.32/0.61).

Load-bearing premise

The load-bearing premise is that the lightweight depth head on frozen geometric features produces dependable per-ray depth posteriors for seeding Gaussian centers; if those posteriors are biased in occluded, distant, or low-light regions, the occupancy output degrades no matter how well the later refinement works.

What would settle it

Run a controlled experiment where the ray-depth prior is corrupted — for example, add a fixed offset to all predicted depths or replace the depth head with a constant-depth predictor — while keeping everything else fixed, and measure SC IoU. A small drop would show that the geometric birth is not the source of the gain; a large drop would confirm it. Alternatively, compare the recall of predicted non-empty depth bins against LiDAR ground truth in the farthest 20 meters of the evaluation region.

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

Share X Bluesky LinkedIn Reddit HN

If this is right

  • Sparse Gaussian occupancy models can match or beat dense voxel and BEV methods without adding Gaussian budget, by investing compute in better initialization and refinement cues.
  • Frozen pretrained geometric and visual features transfer to driving scenes and provide a reusable source of 3D cues for occupancy prediction.
  • Ray-depth-derived centers with spatially balanced sampling improve coverage of occluded and distant regions compared with confidence ranking or farthest-point sampling.
  • Pose-conditioned cross-view aggregation helps resolve per-ray depth ambiguity, as reflected in consistent gains across most semantic classes.
  • Because the decoder is unchanged, the improvement is attributable to how the Gaussians are born and how their refinement features are built, not to a new rendering operator.

Where Pith is reading between the lines

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

  • Since all Gaussian centers are seeded from per-frame ray-depth hypotheses, a natural extension is to aggregate ray hypotheses across multiple frames before sampling; temporal depth priors could further stabilize occupancy in occluded regions, a direction the paper lists as future work.
  • The frozen foundation features set an upper bound: retraining or adapting the geometric backbone on driving-specific data rather than generic imagery is a testable way to push the method further.
  • The ray-compatibility weighting used in cross-view aggregation could be reused in other camera-only 3D tasks, such as monocular depth estimation or 3D detection, as a geometric attention bias; the paper does not claim this.
  • A stress test with systematically biased camera calibration would clarify robustness boundaries, given the large sensitivity to patch-local pose conditioning observed in the ablation study.

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 / 6 minor

Summary. The paper proposes VGOcc, a sparse-Gaussian 3D semantic occupancy model for surround-view driving. It uses frozen VGGT and DINOv2 features to construct 'Visual-Geometric Gaussians': ray-depth hypotheses from a DPT head are lifted to 3D, balanced by voxel-thinned sampling, and paired with DINO attributes to initialize Gaussian queries; a pose-aware feature learning module conditions multi-level foundation tokens on camera/ray geometry and aggregates cross-view features at each refinement stage; the standard Gaussian decoder then refines and renders the occupancy field. On nuScenes with SurroundOcc labels, the authors report 34.07 SC IoU and 21.75 SSC mIoU, exceeding GaussianFormer-2 by 3.51 and 1.73, respectively. Ablations isolate the birth and feature-learning components under a fixed Gaussian budget and decoder.

Significance. If the reported results hold, VGOcc makes a solid contribution to vision-centric occupancy prediction by showing that frozen foundation-model cues can be converted into useful initialization and refinement signals for sparse Gaussian primitives. The design is well motivated, and the ablations are informative; the comparison with GaussianFormer-2 under the same decoder is a good control. The paper is not circular: the headline metrics come from external ground-truth labels and published baselines. However, the empirical contribution is weakened by the absence of the closest Gaussian occupancy baselines (VG3S, VG3T), by single-run results without error bars, and by the lack of code/checkpoint release. These issues need to be addressed before the 'state-of-the-art' claim can be accepted.

major comments (3)
  1. [Quantitative Results / Table 1] The headline SOTA claim is not supported by the baseline set. The Related Work explicitly names VG3S (Yan, Pei, and Shen 2026) and VG3T (Kim and Lee 2025) as sparse Gaussian occupancy methods, and VG3S is described as adapting frozen foundation features for Gaussian occupancy learning — the closest comparator to VGOcc. Neither appears in Table 1. Since the paper's central assertion is that VGOcc achieves state-of-the-art performance, the comparison must include these methods, or provide a specific justification for their exclusion and soften the claim. Please add their published nuScenes/SurroundOcc numbers and update the text accordingly.
  2. [Compared methods / Table 1] The text states that VoxFormer, COTR, and SparseOcc are compared, but these rows are absent from Table 1. The reported baseline table is therefore incomplete relative to the paper's own description. Please either add the missing rows or revise the text. Also clarify which rows use the SurroundOcc dense labels versus the original LiDAR segmentation labels, since the table mixes TPVFormer and TPVFormer†.
  3. [Tables 1 and 2] All quantitative results are reported as single runs with no error bars or significance tests. Given that the central claim is a measured improvement over GaussianFormer-2, the absence of run-to-run variability makes it difficult to assess whether the 1.73 mIoU margin is stable. Please report results over multiple seeds (mean±std) for at least VGOcc and the key baselines, and make code/checkpoints available to reproduce the numbers. If multiple-seed evaluation is not standard for this venue, the paper should state this explicitly.
minor comments (6)
  1. [Experimental Setup] Specify whether the reported results are on the nuScenes validation split, and define the ray-depth posterior supervision: number of depth bins, depth range, and construction of the ray-depth target. This is needed to reproduce Eq. (4).
  2. [Method / Eqs. (4) and (10)] The notation d_b in Eq. (4), and y^j_{c'ik} and δ_k in Eq. (10), are not fully defined. Please add explicit definitions so that the cross-view aggregation is reproducible.
  3. [Figure 1 and typos] Figure 1 contains informal emojis and unclear arrows that are inappropriate for a formal paper; replace with a clean schematic. Also fix typos: 'V enue' in the Table 1 header and 'vison' in the 'Compared methods' paragraph.
  4. [Figure 3 caption] The caption says 'SurroundOcc dataset,' but the experiments are on nuScenes with SurroundOcc labels. Make the wording consistent.
  5. [Efficiency motivation] The paper motivates sparse Gaussians by efficiency but reports no runtime or memory comparison with GaussianFormer-2. A short efficiency table would strengthen the contribution and help quantify the computational overhead of the frozen backbone.
  6. [Ablation B3] The description of 'independent adapters at all levels' versus the default shared adapter could be more precise. Specify what parameter sharing the default uses and why shared adaptation improves feature consistency across the four decoder stages.

Circularity Check

0 steps flagged

No significant circularity: the central SOTA claim is an external nuScenes benchmark comparison, and the method's intermediate outputs are supervised independently of the final metric.

full rationale

The paper's central claim is empirical: VGOcc reports 34.07 SC IoU / 21.75 SSC mIoU on nuScenes in Table 1, evaluated against external SurroundOcc ground truth using the standard mIoU/IoU definitions in Eqs. 15-16. No model component is defined in terms of the final reported metric. The depth posterior p_ray used for Gaussian center birth (Eq. 4) is trained with its own supervision L_pix against ray-depth targets Y_ray, while the final occupancy field is trained with CE + Lovasz against ground truth occupancy (Eqs. 13-14). Thus no fitted parameter is renamed as a prediction, and the Gaussian birth and refinement chain is not self-definitional. The visual and geometric features come from frozen, externally pretrained VGGT and DINOv2; they are not re-derived or re-fit in this paper, so using them as inputs is not circular. The only same-author citation, VGD (Lin et al. 2025), appears in Related Work as context and is not load-bearing: no uniqueness theorem or prior result by these authors is invoked to force VGOcc's design. The absence of VG3S and VG3T from Table 1 is a potential empirical completeness concern, but omitting baselines is a correctness issue, not circularity. The paper also explicitly lists limitations such as reliance on accurate cameras and lack of temporal modeling, which are robustness concerns unrelated to circular reasoning. Overall, the derivation chain is self-contained against external benchmarks and does not reduce to its own inputs.

Axiom & Free-Parameter Ledger

5 free parameters · 5 axioms · 0 invented entities

The paper introduces no new physical entities or forces; the 'Visual-Geometric Gaussians' is a learned representation, not a postulated entity with external falsifiable handles. The load-bearing premises are camera calibration accuracy, transferability of frozen foundation-model features, validity of the SurroundOcc labels, and the benchmark protocol. The free parameters are standard architecture hyperparameters, though their values are often unreported.

free parameters (5)
  • Gaussian budget M = 25.6k (19.2k geometry-guided + 6.4k fallback)
    Hand-chosen capacity; fixed across all experiments and ablations.
  • Number of decoder stages L = 4
    Refinement depth; chosen without reported sensitivity analysis.
  • Auxiliary loss weights λ_pix, λ_aux
    Values are not reported in the paper; hand-set balancing terms in Eq. 13.
  • Voxel size s_v for voxel-thinned sampling
    Controls spatial balance of Gaussian centers; value omitted from the implementation details.
  • Ray-depth bin discretization (d_b)
    Depth posterior bins in Eq. 4 determine candidate centers; bin count/range not specified.
axioms (5)
  • domain assumption Camera calibration (intrinsics K, extrinsics T_{c→e}) is accurate and fixed
    All ray-lifting and cross-view projection use P; the paper lists reliance on accurate cameras as a limitation in the Conclusion.
  • domain assumption Frozen VGGT/DINOv2 features transfer to driving scenes
    The whole pipeline uses these features without fine-tuning; no evidence beyond the reported in-dataset results.
  • domain assumption SurroundOcc 3D occupancy labels are a valid ground truth
    Training and evaluation use SurroundOcc labels; inherent label noise is unexamined.
  • standard math Semantic occupancy metrics (SC IoU, SSC mIoU) are an accepted measure
    Standard benchmark protocol from prior work.
  • domain assumption Sparse Gaussian rendering (GS2Occ) is differentiable and the decoder from GaussianFormer-2 is retained unchanged
    The paper states 'We retain the standard sparse Gaussian refinement and rendering process'; the exact implementation is from prior work without formal justification.

pith-pipeline@v1.3.0-alltime-deepseek · 11890 in / 15129 out tokens · 470670 ms · 2026-08-01T16:08:57.893847+00:00 · methodology

0 comments
Cite this review

Pith. "Pith review of VGOcc: Learning Visual-Geometric Gaussians for Vision-Centric 3D Driving Occupancy Prediction." pith.science (2026). https://pith.science/paper/OOZ7AK5G

@misc{pith2026260718078,
  author       = {Pith},
  title        = {Pith review of: VGOcc: Learning Visual-Geometric Gaussians for Vision-Centric 3D Driving Occupancy Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OOZ7AK5G}},
  note         = {Machine review of arXiv:2607.18078}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Vision-only occupancy prediction requires recovering a semantic 3D occupancy field from calibrated surround-view images, where each view provides observations with ambiguous depth along camera rays. Existing methods have progressed from dense structured representations to sparse Gaussian primitives, improving the efficiency of 3D scene representation. However, Gaussian learning still relies primarily on image domain features, which provide limited explicit geometric information for volumetric reasoning. Our key observation is that effective Gaussian occupancy modeling requires not only sparse primitives, but also richer geometric and semantic learning cues. In this paper, we propose VGOcc, which learns visual and geometric cues from foundation models for Gaussian modeling. VGOcc incorporates these cues into primitive initialization and refinement, yielding a representation termed Visual-Geometric Gaussians tailored to semantic occupancy prediction. Specifically, we propose Visual-Geometric Gaussian Birth to form spatially balanced Gaussian centers from ray depth hypotheses, while visual semantic features initialize primitive attributes. Next, we design Pose-Aware Feature Learning to combine foundation tokens with camera embeddings and calibrated ray information. Features from neighboring views are then aggregated at projected 3D locations for each Gaussian refinement stage. Finally, Gaussian decoder refines birth Gaussians with pose-aware features and renders them into semantic occupancy. Experiments on nuScenes demonstrate that VGOcc achieves state-of-the-art performance in vision-only 3D occupancy prediction. Codes will be available at https://github.com/JHLin42in/VGOcc.

Figures

Figures reproduced from arXiv: 2607.18078 by Junhong Lin, Kangli Wang, Wei Gao, Xianda Guo, Xiaoyu Liang, Yanlun Peng, Yuqi Ye.

Figure 1
Figure 1. Figure 1: From structured lifting to Visual-Geometric Gaus￾sians. BEV, tri-plane, and voxel methods establish structured 3D representations, while sparse Gaussians reduce spatial redundancy. VGOcc learns Visual-Geometric Gaussians by grounding their initialization and refinement in complemen￾tary visual and geometric cues. structure. Each image location defines a ray but does not re￾veal the occupied depth or struct… view at source ↗
Figure 2
Figure 2. Figure 2: Overall architecture of VGOcc. Frozen VGGT provides complementary visual and geometric tokens. Visual [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Qualitative results of our method and GaussianFormer-2 on SurroundOcc dataset. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative results of ablation studies. Red and Green rectangular boxes highlight the differences. [PITH_FULL_IMAGE:figures/full_fig_p007_4.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

50 extracted references · 7 linked inside Pith

  1. [1]

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

    nuScenes: A Multimodal Dataset for Autonomous Driving , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  2. [2]

    Proceedings of the European Conference on Computer Vision (ECCV) , pages=

    BEVFormer: Learning Bird's-Eye-View Representation from Multi-Camera Images via Spatiotemporal Transformers , author=. Proceedings of the European Conference on Computer Vision (ECCV) , pages=

  3. [3]

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

    Tri-Perspective View for Vision-Based 3D Semantic Occupancy Prediction , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  4. [4]

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

    SurroundOcc: Multi-Camera 3D Occupancy Prediction for Autonomous Driving , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  5. [5]

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

    OccFormer: Dual-Path Transformer for Vision-Based 3D Semantic Occupancy Prediction , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  6. [6]

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

    VoxFormer: Sparse Voxel Transformer for Camera-Based 3D Semantic Scene Completion , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  7. [7]

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

    COTR: Compact Occupancy TRansformer for Vision-Based 3D Occupancy Prediction , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  8. [8]

    Advances in Neural Information Processing Systems, Datasets and Benchmarks Track , year=

    Occ3D: A Large-Scale 3D Occupancy Prediction Benchmark for Autonomous Driving , author=. Advances in Neural Information Processing Systems, Datasets and Benchmarks Track , year=

  9. [9]

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

    OpenOccupancy: A Large Scale Benchmark for Surrounding Semantic Occupancy Perception , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  10. [10]

    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=

  11. [11]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Regulating Intermediate 3D Features for Vision-Centric Autonomous Driving , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  12. [12]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    ProtoOcc: Accurate, Efficient 3D Occupancy Prediction Using Dual Branch Encoder-Prototype Query Decoder , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  13. [13]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Achieving Speed-Accuracy Balance in Vision-Based 3D Occupancy Prediction via Geometric-Semantic Disentanglement , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  14. [14]

    ACM Transactions on Graphics , volume=

    3D Gaussian Splatting for Real-Time Radiance Field Rendering , author=. ACM Transactions on Graphics , volume=

  15. [15]

    Proceedings of the European Conference on Computer Vision (ECCV) , year=

    GaussianFormer: Scene as Gaussians for Vision-Based 3D Semantic Occupancy Prediction , author=. Proceedings of the European Conference on Computer Vision (ECCV) , year=

  16. [16]

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

    GaussianFormer-2: Probabilistic Gaussian Superposition for Efficient 3D Occupancy Prediction , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  17. [17]

    arXiv preprint arXiv:2505.10685 , year=

    GaussianFormer3D: Multi-Modal Gaussian-Based Semantic Occupancy Prediction with 3D Deformable Attention , author=. arXiv preprint arXiv:2505.10685 , year=

    arXiv

  18. [18]

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

    SelfOcc: Self-Supervised Vision-Based 3D Occupancy Prediction , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  19. [19]

    arXiv preprint arXiv:2312.09243 , year=

    OccNeRF: Advancing 3D Occupancy Prediction in LiDAR-Free Environments , author=. arXiv preprint arXiv:2312.09243 , year=

    Pith/arXiv arXiv

  20. [20]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    ViPOcc: Leveraging Visual Priors from Vision Foundation Models for Single-View 3D Occupancy Prediction , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  21. [21]

    Transactions on Machine Learning Research , year=

    DINOv2: Learning Robust Visual Features without Supervision , author=. Transactions on Machine Learning Research , year=

  22. [22]

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

    Segment Anything , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  23. [23]

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

    Vision Transformers for Dense Prediction , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  24. [24]

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

    Feature Pyramid Networks for Object Detection , author=. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages=

  25. [25]

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

    DUSt3R: Geometric 3D Vision Made Easy , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  26. [26]

    Proceedings of the European Conference on Computer Vision (ECCV) , pages=

    Grounding Image Matching in 3D with MASt3R , author=. Proceedings of the European Conference on Computer Vision (ECCV) , pages=

  27. [27]

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

    VGGT: Visual Geometry Grounded Transformer , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  28. [28]

    arXiv preprint arXiv:2512.05988 , year=

    VG3T: Visual Geometry Grounded Gaussian Transformer , author=. arXiv preprint arXiv:2512.05988 , year=

    arXiv

  29. [29]

    arXiv preprint arXiv:2603.06210 , year=

    VG3S: Visual Geometry Grounded Gaussian Splatting for Semantic Occupancy Prediction , author=. arXiv preprint arXiv:2603.06210 , year=

    Pith/arXiv arXiv

  30. [30]

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

    The Lovasz-Softmax Loss: A Tractable Surrogate for the Optimization of the Intersection-over-Union Measure in Neural Networks , author=. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages=

  31. [31]

    Proceedings of the European Conference on Computer Vision (ECCV) , year=

    Fully Sparse 3D Occupancy Prediction , author=. Proceedings of the European Conference on Computer Vision (ECCV) , year=

  32. [32]

    arXiv preprint arXiv:2510.19578 , year=

    VGD: Visual Geometry Gaussian Splatting for Feed-Forward Surround-view Driving Reconstruction , author=. arXiv preprint arXiv:2510.19578 , year=

    arXiv

  33. [33]

    arXiv preprint arXiv:2512.03004 , year=

    DGGT: Feedforward 4D Reconstruction of Dynamic Driving Scenes using Unposed Images , author=. arXiv preprint arXiv:2512.03004 , year=

    arXiv

  34. [34]

    arXiv preprint arXiv:2511.22264 , year=

    DriveVGGT: Calibration-Constrained Visual Geometry Transformers for Multi-Camera Autonomous Driving , author=. arXiv preprint arXiv:2511.22264 , year=

    arXiv

  35. [35]

    arXiv preprint arXiv:2512.16919 , year=

    DVGT: Driving Visual Geometry Transformer , author=. arXiv preprint arXiv:2512.16919 , year=

    Pith/arXiv arXiv

  36. [36]

    arXiv preprint arXiv:2604.00813 , year=

    DVGT-2: Vision-Geometry-Action Model for Autonomous Driving at Scale , author=. arXiv preprint arXiv:2604.00813 , year=

    Pith/arXiv arXiv

  37. [37]

    arXiv preprint arXiv:2601.06474 , year=

    SparseOccVLA: Bridging Occupancy and Vision-Language Models via Sparse Queries for Unified 4D Scene Understanding and Planning , author=. arXiv preprint arXiv:2601.06474 , year=

    arXiv

  38. [38]

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

    VGGDrive: Empowering Vision-Language Models with Cross-View Geometric Grounding for Autonomous Driving , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

  39. [39]

    Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    UniOcc: A Unified Benchmark for Occupancy Forecasting and Prediction in Autonomous Driving , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year=

  40. [40]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Driving in the Occupancy World: Vision-Centric 4D Occupancy Forecasting and Planning via World Models for Autonomous Driving , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  41. [41]

    International Conference on Learning Representations , year=

    PreWorld: Semi-Supervised Vision-Centric 3D Occupancy World Model for Autonomous Driving , author=. International Conference on Learning Representations , year=

  42. [42]

    CVPR Workshop on End-to-End Autonomous Driving , year =

    FB-OCC: 3D Occupancy Prediction based on Forward-Backward View Transformation , author =. CVPR Workshop on End-to-End Autonomous Driving , year =

  43. [43]

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

    MonoScene: Monocular 3D Semantic Scene Completion , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  44. [44]

    Proceedings of the European Conference on Computer Vision (ECCV) , pages=

    Atlas: End-to-End 3D Scene Reconstruction from Posed Images , author=. Proceedings of the European Conference on Computer Vision (ECCV) , pages=. 2020 , organization=

    2020

  45. [45]

    arXiv preprint arXiv:2606.22971 , year=

    Humanoid-OmniOcc: Stereo-Based Full-View Occupancy Dataset for Embodied AI , author=. arXiv preprint arXiv:2606.22971 , year=

    Pith/arXiv arXiv

  46. [46]

    Conference on Neural Information Processing Systems (NeurIPS) , year=

    SURDS: Benchmarking Spatial Understanding and Reasoning in Driving Scenarios with Vision Language Models , author=. Conference on Neural Information Processing Systems (NeurIPS) , year=

  47. [47]

    Proceedings of the European Conference on Computer Vision (ECCV) , year=

    OmniNWM: Omniscient Driving Navigation World Models , author=. Proceedings of the European Conference on Computer Vision (ECCV) , year=

  48. [48]

    arXiv preprint arXiv:2603.28116 , year=

    AutoDrive-P ^3 : Unified Chain of Perception--Prediction--Planning Thought via Reinforcement Fine-Tuning , author=. arXiv preprint arXiv:2603.28116 , year=

    arXiv

  49. [49]

    arXiv preprint arXiv:2607.05543 , year=

    GEM-Occ: From Visual Geometry Evidence to Embodied Semantic Occupancy Memory , author=. arXiv preprint arXiv:2607.05543 , year=

    Pith/arXiv arXiv

  50. [50]

    arXiv preprint arXiv:2607.01633 , year=

    Bridging 3D Gaussians and Semantic Occupancy for Comprehensive Open-Vocabulary Scene Understanding from Unposed Images , author=. arXiv preprint arXiv:2607.01633 , year=

    Pith/arXiv arXiv