Pith. sign in

REVIEW 3 major objections 5 minor 4 cited by

CRUISE: Cooperative Reconstruction and Editing in V2X Scenarios using Gaussian Splatting

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read CRUISE claims that a Gaussian-splatting V2X simulator can generate synchronized ego-infrastructure data whose addition to real training data improves 3D detection and tracking.

desk verdict First GS-based V2X simulator with credible reconstruction, but the augmentation gain is only shown on the same four intersections used for generation—so the generalization claim needs a held-out test. read the letter →

arxiv 2507.18473 v1 pith:PJHOVCVP submitted 2025-07-24 cs.CV

Haoran Xu , Saining Zhang , Peishuo Li , Baijun Ye , Xiaoxue Chen , Huan-ang Gao , Jv Zheng , Xiaowei Song
show 10 more authors
Ziqiao Peng Run Miao Jinrang Jia Yifeng Shi Guangqi Yi Hang Zhao Hao Tang Hongyang Li Kaicheng Yu Hao Zhao
This is my paper
classification cs.CV
keywords V2XGaussianSplattingautonomousdrivingsimulationdataaugmentation3Dobjectdetectioncooperativeperceptionsceneeditingsynthetic
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

CRUISE is a simulation framework that reconstructs real V2X driving scenes as decomposed Gaussians, separating moving vehicles from the static street, and then edits those scenes by inserting generated 3D vehicle assets along language-model-drafted trajectories. From the edited scenes it renders synchronized images from both the ego vehicle and roadside infrastructure, complete with 3D annotation boxes. The paper's central quantitative claim is that training 3D detectors and trackers on a mix of real data and this generated data improves performance over real data alone, in ego, infrastructure, and cooperative views. If that holds, generated V2X data becomes a practical augmentation for perception models without new real-world capture.

What carries the argument

The load-bearing object is the decomposed Gaussian representation: dynamic traffic participants are 4D Gaussians with tracked poses and appearance variation, while the static street is a separate 3D Gaussian background. An ego-mask prevents the ego-car's bounding box from swallowing background geometry, scale and ratio losses flatten Gaussians toward surfaces, and appearance decoupling removes uneven illumination. This decomposition is what lets the pipeline edit scenes by inserting vehicle assets and re-render from any viewpoint.

What would settle it

Hold out several V2X-Seq intersections entirely from reconstruction and generation, train detectors on real plus generated data from the seen intersections, and test on the held-out intersections; if the mixed training no longer beats real-only training, the generalization claim fails.

Watch

Extended reading notes

Core claim

The discovery, stated in the paper's own terms, is that a Gaussian-splatting-based V2X simulator can serve as a closed-loop data generator: reconstructing real intersections, editing them with new vehicles and trajectories, and re-rendering both ego and infrastructure views with perfectly synchronized timestamps and precise boxes. On V2X-Seq, mixing generated frames with real frames outperforms real-only training for ego-view detection (moderate AP3D rising from 37.69 to 41.05), infrastructure detection (50.16 to 54.26), cooperative detection (AP3D 14.79 to 15.91), and cooperative tracking (MOTA 21.83 to 25.52). The paper argues the generated data provides out-of-distribution samples and cleaner tracking labels, which is why the hybrid strategy wins.

Load-bearing premise

The reported improvements are measured on test sequences from the same four intersections used for reconstruction and generation, so the load-bearing premise is that generated data also helps on scenes the simulator has not seen.

Editorial extensions

If this is right

  • Generated data alone matches or exceeds real data for single-view and cooperative detection, and exceeds real data for tracking (MOTA 25.03 vs 21.83).
  • Combining real and generated data improves performance across ego (AP3D moderate 41.05 vs 37.69), infrastructure (54.26 vs 50.16), and cooperative (15.91 vs 14.79) 3D detection.
  • CRUISE produces perfectly synchronized ego and infrastructure frames, eliminating temporal misalignment common in real V2X datasets.
  • The framework can synthesize corner cases, such as an occluded vehicle visible only to one view, useful for testing V2X safety.

Reading between the lines

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

  • A direct testable extension is measuring gains on intersections outside the reconstruction set; the method's value beyond seen static scenes is currently unproven.
  • Because the pipeline renders camera images only and does not simulate realistic LiDAR point clouds, its augmentation benefit is likely limited to camera-based perception models.
  • The language-model-drafted trajectories may not match realistic traffic flow; comparing against a physics- or rule-based traffic simulator would reveal whether this choice matters.
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

3 major / 5 minor

Summary. CRUISE is a Gaussian-splatting-based framework that reconstructs real V2X driving scenes from V2X-Seq, decomposes them into static backgrounds and dynamic vehicle Gaussians, edits the scenes by inserting TRELLIS-generated vehicle assets along GPT-4o-drafted trajectories, and renders synchronized ego-vehicle and infrastructure images with 3D annotations. The authors evaluate reconstructed scene quality against several GS baselines, ablate four reconstruction modules, and use the generated images as training data for MonoLSS, BevHeight, and ImVoxelNet on ego, infrastructure, and cooperative 3D detection and tracking. They report that combining real and generated data improves over real-only training and show qualitative corner cases.

Significance. If the main quantitative claim holds, CRUISE would be a useful step toward editable, reconstruction-based V2X data augmentation, combining an explicit Gaussian representation with LLM-guided traffic editing and synchronized multi-view rendering. The paper is honest about limitations (unrealistic LiDAR, rain artifacts) and includes quantitative ablations. The reconstruction component largely inherits Street Gaussians, so the novel contribution is the V2X generation pipeline; the current evaluation is not yet sufficient to establish the claimed downstream gains as a general data-augmentation result.

major comments (3)
  1. [Sec. V-E, Tables III-V] The comparison underpinning the central claim is not fully specified. The text says the Real and Gen training sets each contain 1000 frames and that the "Real + Gen" model is trained "for half the training iterations," but it never states whether iterations, epochs, batch sizes, learning-rate schedules, and total numbers of real vs. generated samples are matched across conditions. If "half the training iterations" means half as many epochs over a doubled dataset, Real+Gen consumes the same total number of samples as Real but only half as many real samples; if it means half of Real's iteration count with the same batch size, Real+Gen consumes fewer total samples. These interpretations have opposite implications for whether the gains in Tables III-V reflect generated-data diversity or training budget. Please report the exact optimization schedule for every row in Tables III-V.
  2. [Sec. V-A and V-E] The test set is drawn only from the same four intersections used for reconstruction and generation, and all downstream detectors are image-based. Because the generated images are rendered from reconstructed backgrounds of those same intersections, a Real+Gen model could improve by exploiting scene-specific visual cues rather than by learning a generalizable augmentation. Since V2X-Seq contains six intersections, a held-out-intersection evaluation is feasible and should be reported; at a minimum, the paper should show per-intersection results or a cross-intersection transfer experiment before claiming that CRUISE improves detection and tracking as a general data-generation method.
  3. [Tables III-V] The reported improvements are small (e.g., cooperative AP3D rises from 14.79 to 15.91 in Table V) and appear to come from a single training run per condition. Without error bars or multiple seeds, it is unclear whether the differences are statistically meaningful. Please report mean and standard deviation over at least three random seeds, or provide other evidence that the differences are not within training noise.
minor comments (5)
  1. [Sec. III, Eq. (2)] Lratio = max(1, s1/s2) - 1 is redundant because s1 is defined as the longest and s2 the second-longest scaling factor, so s1/s2 >= 1; please clarify the ordering convention and simplify if intended.
  2. [Sec. V-B] The downstream task training details (optimizer, learning rate, batch size, data augmentation) are omitted; please include them for reproducibility.
  3. [Table I] The row "Street-GS(Ours) [28]" is ambiguous; specify whether this is vanilla Street Gaussians or Street Gaussians with CRUISE's modifications (ego-mask, normal loss, scale/ratio losses, appearance decoupling), and add a vanilla Street Gaussians baseline if the latter.
  4. [Sec. VII] "Although CRUISE could generate highly accurate and realistic V2X data" should be "Although CRUISE can generate..." for grammatical clarity.
  5. [Sec. IV-B, Fig. 3] The prompt and decoding parameters for GPT-4o trajectory generation are not described; a brief reproducibility note would help.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; CRUISE's empirical claims are benchmarked against real V2X-Seq data, and its component choices are adopted from external prior work.

full rationale

CRUISE does not derive its headline result from its own fitted quantities. The central claim—that Real+Gen training improves 3D detection and tracking over Real-only training (Tables III–V)—is an empirical comparison on the V2X-Seq benchmark, with generated data produced by a pipeline (reconstruction via Street Gaussians [28], asset creation via TRELLIS [77], trajectory drafting via GPT-4o) that is not tuned to optimize the downstream detector metrics on the test sequences. The reconstruction objective (Eq. 3) is a standard photometric, geometric, and semantic loss on the source sequences, not a function of downstream AP or MOTA, so the Table III–V gains are not forced by construction. The adopted base method, Street Gaussians [28], is external prior work rather than a self-citation chain; the same-intersection test design (Sec. V-E: 8 test sequences from the 4 reconstruction intersections) limits generalization evidence but does not make the comparison circular. No load-bearing uniqueness theorem, ansatz, or renamed known result was found; the stated limitations (Sec. VII) are honest scope restrictions rather than circular moves. Therefore score 0.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The paper's central claim rests on the reliability of external components (Street Gaussians, TRELLIS, GPT-4o, Grounded SAM 2, StableNormal) and on the choice of dataset splits. The only quantities fitted to data are the loss weights and training lengths, none of which are reported with values. No new physical or mathematical entities are postulated.

free parameters (4)
  • Loss weights lambda_1..lambda_7 = not reported
    The reconstruction loss in Eq. (3) combines color, depth, normal, sky, semantic, scale, ratio, and regularization terms; the relative weights are chosen by hand and not specified, and they directly affect reconstruction quality that downstream generation depends on.
  • Reconstruction iterations = 50000
    Chosen for Street Gaussians training; not varied.
  • Number of generated frames = 1000
    Arbitrary training-set size for downstream detectors; not justified or ablated.
  • Detection training epochs = 150/100/24
    Chosen per detector (MonoLSS, BEVHeight, ImVoxelNet); not ablated.
assumptions (5)
  • domain assumption Street Gaussians [28] reliably reconstructs dynamic urban scenes from posed images and LiDAR.
    The entire pipeline adopts Street Gaussians as the reconstruction backbone (Sec. IV-A); if this method fails on V2X dual-view setups, the generated data inherits the failure.
  • domain assumption V2X-Seq tracking annotations are precise enough to supervise foreground/background decomposition.
    Dynamic vehicles are separated using annotation boxes (Sec. IV-A); the paper later claims V2X-Seq annotations contain errors, which makes the use of these same boxes as ground truth for decomposition and for the 'gen' labels internally inconsistent.
  • ad hoc to paper GPT-4o generates physically plausible vehicle trajectories from a simplified vector map.
    Scene editing inserts vehicles along GPT-4o-suggested paths (Sec. IV-B); no collision checking, traffic rule validation, or human evaluation is reported.
  • domain assumption TRELLIS produces 3D Gaussian assets whose appearance and geometry are realistic enough for downstream detection training.
    Internet images are converted to assets with TRELLIS (Sec. IV-B); asset quality is never quantitatively evaluated.
  • ad hoc to paper Test sequences from the same four intersections provide an unbiased measure of generated-data utility.
    Sec. V-E constructs the test set from the same intersections used for reconstruction, so background familiarity may inflate gains.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CRUISE: Cooperative Reconstruction and Editing in V2X Scenarios using Gaussian Splatting." pith.science (2026). https://pith.science/paper/PJHOVCVP

@misc{pith2026250718473,
  author       = {Pith},
  title        = {Pith review of: CRUISE: Cooperative Reconstruction and Editing in V2X Scenarios using Gaussian Splatting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PJHOVCVP}},
  note         = {Machine review of arXiv:2507.18473}
}
read the original abstract

Vehicle-to-everything (V2X) communication plays a crucial role in autonomous driving, enabling cooperation between vehicles and infrastructure. While simulation has significantly contributed to various autonomous driving tasks, its potential for data generation and augmentation in V2X scenarios remains underexplored. In this paper, we introduce CRUISE, a comprehensive reconstruction-and-synthesis framework designed for V2X driving environments. CRUISE employs decomposed Gaussian Splatting to accurately reconstruct real-world scenes while supporting flexible editing. By decomposing dynamic traffic participants into editable Gaussian representations, CRUISE allows for seamless modification and augmentation of driving scenes. Furthermore, the framework renders images from both ego-vehicle and infrastructure views, enabling large-scale V2X dataset augmentation for training and evaluation. Our experimental results demonstrate that: 1) CRUISE reconstructs real-world V2X driving scenes with high fidelity; 2) using CRUISE improves 3D detection across ego-vehicle, infrastructure, and cooperative views, as well as cooperative 3D tracking on the V2X-Seq benchmark; and 3) CRUISE effectively generates challenging corner cases.

Figures

Figures reproduced from arXiv: 2507.18473 by the authors.

Figure 1
Figure 1. On the left is the V2X setup, showing captures from both the ego [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The workflow of CRUISE. The data layer processes the V2X-Seq dataset into suitable format for further reconstruction. The processed data is [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The pipeline of scene editing. By collecting basic vehicle informa [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Data generated by CRUISE. Up-to-down: Infrastructure views; Corresponding Ego-vehicle views. The red arrow points to the position of the [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Qualitative results of cooperative 3D detection on the original V2X-Seq dataset. The upper row shows infrastructure views, while the lower row [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: The visualization of the corner case: A demonstration of the vehicle [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]

Discussion (0). Sign in to comment.

Forward citations

Cited by 4 Pith papers

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

  1. Real2Sim: A Physics-driven and Editable Gaussian Splatting Framework for Autonomous Driving Scenes

    cs.CV 2026-05 unverdicted novelty 6.0 of 10

    Real2Sim reconstructs editable dynamic driving scenes as temporally continuous Gaussians integrated with a differentiable MPM physics solver for high-fidelity simulation of interactions and collisions.

  2. You Only Gaussian Once: Controllable 3D Gaussian Splatting for Ultra-Densely Sampled Scenes

    cs.CV 2026-04 unverdicted novelty 6.0 of 10

    YOGO reformulates stochastic 3D Gaussian Splatting into a deterministic budget-aware system and supplies an ultra-dense dataset to enforce physical fidelity over viewpoint interpolation.

  3. You Only Gaussian Once: Controllable 3D Gaussian Splatting for Ultra-Densely Sampled Scenes

    cs.CV 2026-04 conditional novelty 6.0 of 10

    YOGO delivers deterministic budget-controlled 3D Gaussian Splatting that matches or exceeds prior methods on a new ultra-dense multi-sensor indoor benchmark while keeping primitive counts strictly fixed.

  4. You Only Gaussian Once: Controllable 3D Gaussian Splatting for Ultra-Densely Sampled Scenes

    cs.CV 2026-04 reject novelty 6.0 of 10

    YOGO enforces a fixed Gaussian budget during training via a deterministic controller, and the dense Immersion dataset shifts evaluation from sparse-view interpolation to physical fidelity.

Reference graph

Works this paper leans on

83 extracted references · 61 canonical work pages · cited by 2 Pith papers

  1. [1]

    Unifying appearance codes and bilateral grids for driving scene gaussian splatting,

    N. Wang, Y . Chen, L. Xiao, W. Xiao, B. Li, Z. Chen, C. Ye, S. Xu, S. Zhang, Z. Yan et al., “Unifying appearance codes and bilateral grids for driving scene gaussian splatting,”arXiv preprint arXiv:2506.05280, 2025

    arXiv 2025

  2. [2]

    Controllable Radar Simulation with Waveform Parameter Embedding

    W. Xiao, H. Huang, C. Zhong, Y . Lin, N. Wang, X. Chen, Z. Chen, S. Zhang, S. Yang, P. Merriaux et al. , “Simulate any radar: Attribute- controllable radar simulation via waveform parameter embedding,” arXiv preprint arXiv:2506.03134 , 2025

    work page Pith review arXiv 2025

  3. [3]

    Challenger: Affordable adversarial driving video generation,

    Z. Xu, B. Li, H.-a. Gao, M. Gao, Y . Chen, M. Liu, C. Yan, H. Zhao, S. Feng, and H. Zhao, “Challenger: Affordable adversarial driving video generation,” arXiv preprint arXiv:2505.15880 , 2025

    arXiv 2025

  4. [4]

    Dist-4d: Disentangled spatiotemporal diffusion with metric depth for 4d driving scene generation,

    J. Guo, Y . Ding, X. Chen, S. Chen, B. Li, Y . Zou, X. Lyu, F. Tan, X. Qi, Z. Li et al. , “Dist-4d: Disentangled spatiotemporal diffusion with metric depth for 4d driving scene generation,” arXiv preprint arXiv:2503.15208, 2025

    arXiv 2025

  5. [5]

    Avd2: Accident video diffusion for accident video description,

    C. Li, K. Zhou, T. Liu, Y . Wang, M. Zhuang, H.-a. Gao, B. Jin, and H. Zhao, “Avd2: Accident video diffusion for accident video description,” arXiv preprint arXiv:2502.14801 , 2025

    arXiv 2025

  6. [6]

    Uniscene: Unified occupancy-centric driving scene generation,

    B. Li, J. Guo, H. Liu, Y . Zou, Y . Ding, X. Chen, H. Zhu, F. Tan, C. Zhang, T. Wang et al. , “Uniscene: Unified occupancy-centric driving scene generation,” in Proceedings of the Computer Vision and Pattern Recognition Conference , 2025, pp. 11 971–11 981

    2025

  7. [7]

    Scp- diff: Spatial-categorical joint prior for diffusion based semantic image synthesis,

    H.-a. Gao, M. Gao, J. Li, W. Li, R. Zhi, H. Tang, and H. Zhao, “Scp- diff: Spatial-categorical joint prior for diffusion based semantic image synthesis,” in European Conference on Computer Vision . Springer, 2024, pp. 37–54

    2024

  8. [8]

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

    B. Mildenhall, P. P. Srinivasan, M. Tancik, J. T. Barron, R. Ramamoor- thi, and R. Ng, “Nerf: Representing scenes as neural radiance fields for view synthesis,” in ECCV, 2020

    2020

Show all 83 references
  1. [9]

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

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

    2021

  2. [10]

    Slimmerf: Slimmable radiance fields,

    S. Yuan and H. Zhao, “Slimmerf: Slimmable radiance fields,” in 2024 International Conference on 3D Vision (3DV) . IEEE, 2024, pp. 64– 74

    2024

  3. [11]

    Rip-nerf: Anti-aliasing radiance fields with ripmap-encoded platonic solids,

    J. Liu, W. Hu, Z. Yang, J. Chen, G. Wang, X. Chen, Y . Cai, H.- a. Gao, and H. Zhao, “Rip-nerf: Anti-aliasing radiance fields with ripmap-encoded platonic solids,” in ACM SIGGRAPH 2024 Confer- ence Papers, 2024, pp. 1–11

    2024

  4. [12]

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

    B. Kerbl, G. Kopanas, T. Leimk ¨uhler, and G. Drettakis, “3d gaussian splatting for real-time radiance field rendering,” TOG, vol. 42, no. 4, July 2023

    2023

  5. [13]

    Block-nerf: Scalable large scene neural view synthesis,

    M. Tancik, V . Casser, X. Yan, S. Pradhan, B. Mildenhall, P. P. Srinivasan, J. T. Barron, and H. Kretzschmar, “Block-nerf: Scalable large scene neural view synthesis,” in CVPR, 2022

    2022

  6. [14]

    Nerf in the wild: Neural radiance fields for unconstrained photo collections,

    R. Martin-Brualla, N. Radwan, M. S. Sajjadi, J. T. Barron, A. Doso- vitskiy, and D. Duckworth, “Nerf in the wild: Neural radiance fields for unconstrained photo collections,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2021, pp. 7210–7219

    2021

  7. [15]

    Urban radiance fields,

    K. Rematas, A. Liu, P. P. Srinivasan, J. T. Barron, A. Tagliasacchi, T. Funkhouser, and V . Ferrari, “Urban radiance fields,” inCVPR, 2022

    2022

  8. [16]

    Streetsurf: Extending multi-view implicit surface reconstruction to street views,

    J. Guo, N. Deng, X. Li, Y . Bai, B. Shi, C. Wang, C. Ding, D. Wang, and Y . Li, “Streetsurf: Extending multi-view implicit surface reconstruction to street views,” arXiv preprint arXiv:2306.04988 , 2023

    2023 arXiv

  9. [17]

    Oasim: an open and adaptive simulator based on neural rendering for autonomous driving,

    G. Yan, J. Pi, J. Guo, Z. Luo, M. Dou, N. Deng, Q. Huang, D. Fu, L. Wen, P. Cai et al. , “Oasim: an open and adaptive simulator based on neural rendering for autonomous driving,” arXiv preprint arXiv:2402.03830, 2024

    2024 arXiv

  10. [18]

    Drone-assisted road gaussian splatting with cross-view uncertainty,

    S. Zhang, B. Ye, X. Chen, Y . Chen, Z. Zhang, C. Peng, Y . Shi, and H. Zhao, “Drone-assisted road gaussian splatting with cross-view uncertainty,” arXiv preprint arXiv:2408.15242 , 2024

    2024 arXiv

  11. [19]

    Emernerf: Emergent spatial-temporal scene decomposition via self-supervision,

    J. Yang, B. Ivanovic, O. Litany, X. Weng, S. W. Kim, B. Li, T. Che, D. Xu, S. Fidler, M. Pavone, and Y . Wang, “Emernerf: Emergent spatial-temporal scene decomposition via self-supervision,” in ICLR, 2024

    2024

  12. [20]

    Suds: Scalable urban dynamic scenes,

    H. Turki, J. Y . Zhang, F. Ferroni, and D. Ramanan, “Suds: Scalable urban dynamic scenes,” in CVPR, 2023

    2023

  13. [21]

    Neural scene graphs for dynamic scenes,

    J. Ost, F. Mannan, N. Thuerey, J. Knodt, and F. Heide, “Neural scene graphs for dynamic scenes,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2021, pp. 2856–2865

    2021

  14. [22]

    Panoptic neural fields: A semantic object-aware neural scene representation,

    A. Kundu, K. Genova, X. Yin, A. Fathi, C. Pantofaru, L. J. Guibas, A. Tagliasacchi, F. Dellaert, and T. Funkhouser, “Panoptic neural fields: A semantic object-aware neural scene representation,” in Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni...

    2022

  15. [23]

    Unisim: A neural closed-loop sensor simulator,

    Z. Yang, Y . Chen, J. Wang, S. Manivasagam, W.-C. Ma, A. J. Yang, and R. Urtasun, “Unisim: A neural closed-loop sensor simulator,” in CVPR, 2023

    2023

  16. [24]

    Mars: An instance-aware, modular and realistic simulator for autonomous driving,

    Z. Wu, T. Liu, L. Luo, Z. Zhong, J. Chen, H. Xiao, C. Hou, H. Lou, Y . Chen, R. Yang, Y . Huang, X. Ye, Z. Yan, Y . Shi, Y . Liao, and H. Zhao, “Mars: An instance-aware, modular and realistic simulator for autonomous driving,” in CICAI, 2023

    2023

  17. [25]

    Neurad: Neural rendering for autonomous driving,

    A. Tonderski, C. Lindstr ¨om, G. Hess, W. Ljungbergh, L. Svensson, and C. Petersson, “Neurad: Neural rendering for autonomous driving,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 14 895–14 904

    2024

  18. [26]

    Multi-level neural scene graphs for dynamic urban environments,

    T. Fischer, L. Porzi, S. R. Bulo, M. Pollefeys, and P. Kontschieder, “Multi-level neural scene graphs for dynamic urban environments,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 21 125–21 135

    2024

  19. [27]

    Driv- inggaussian: Composite gaussian splatting for surrounding dynamic autonomous driving scenes,

    X. Zhou, Z. Lin, X. Shan, Y . Wang, D. Sun, and M.-H. Yang, “Driv- inggaussian: Composite gaussian splatting for surrounding dynamic autonomous driving scenes,” arXiv preprint arXiv:2312.07920 , 2023

    2023 arXiv

  20. [28]

    Street gaussians for modeling dynamic urban scenes,

    Y . Yan, H. Lin, C. Zhou, W. Wang, H. Sun, K. Zhan, X. Lang, X. Zhou, and S. Peng, “Street gaussians for modeling dynamic urban scenes,” arXiv preprint arXiv:2401.01339 , 2024

    2024 arXiv

  21. [29]

    S 3 gaussian: Self-supervised street gaus- sians for autonomous driving,

    N. Huang, X. Wei, W. Zheng, P. An, M. Lu, W. Zhan, M. Tomizuka, K. Keutzer, and S. Zhang, “S 3 gaussian: Self-supervised street gaus- sians for autonomous driving,”arXiv preprint arXiv:2405.20323, 2024

    2024 arXiv

  22. [30]

    Autonomous vehicle platoons in urban road networks: A joint distributed reinforcement learning and model predictive control approach,

    L. D’Alfonso, F. Giannini, G. Franz `e, G. Fedele, F. Pupo, and G. Fortino, “Autonomous vehicle platoons in urban road networks: A joint distributed reinforcement learning and model predictive control approach,” IEEE/CAA Journal of Automatica Sinica , vol. 11, no. 1, pp. 141–156, 2024

    2024

  23. [31]

    An autonomous vehicle group cooperation model in an urban scene,

    G. Yuan, J. Cheng, M. Zhou, S. Cheng, S. Gao, C. Jiang, and A. Abusorrah, “An autonomous vehicle group cooperation model in an urban scene,” IEEE Transactions on Intelligent Transportation Systems, vol. 24, no. 12, pp. 13 852–13 862, 2023

    2023

  24. [32]

    V2vnet: Vehicle-to-vehicle communication for joint per- ception and prediction,

    T.-H. Wang, S. Manivasagam, M. Liang, B. Yang, W. Zeng, and R. Urtasun, “V2vnet: Vehicle-to-vehicle communication for joint per- ception and prediction,” in European Conference on Computer Vision . Springer, 2020, pp. 605–621

    2020

  25. [33]

    Collaboration helps camera overtake lidar in 3D detection,

    Y . Hu, Y . Lu, R. Xu, W. Xie, S. Chen, and Y . Wang, “Collaboration helps camera overtake lidar in 3D detection,” 2023 IEEE/CVF Con- ference on Computer Vision and Pattern Recognition (CVPR) , 2023

    2023

  26. [34]

    Flow-based feature fusion for vehicle-infrastructure cooperative 3D object detection,

    H. Yu, Y . Tang, E. Xie, J. Mao, P. Luo, and Z. Nie, “Flow-based feature fusion for vehicle-infrastructure cooperative 3D object detection,” in Advances in Neural Information Processing Systems (NeurIPS) , 2023

    2023

  27. [35]

    Umc: A unified bandwidth-efficient and multi-resolution based collaborative perception framework,

    W. Tianhang, C. Guang, C. Kai, L. Zhengfa, Z. Bo, K. Alois, and J. Changjun, “Umc: A unified bandwidth-efficient and multi-resolution based collaborative perception framework,” in Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , 2023

    2023

  28. [36]

    Distributed data-sharing consensus in cooperative perception of autonomous vehicles,

    C. Qiu, S. Yadav, A. Squicciarini, Q. Yang, S. Fu, J. Zhao, and C. Xu, “Distributed data-sharing consensus in cooperative perception of autonomous vehicles,” in 2022 IEEE 42nd International Conference on Distributed Computing Systems (ICDCS) . IEEE, 2022, pp. 1212– 1222

    2022

  29. [37]

    V2x-seq: A large- scale sequential dataset for vehicle-infrastructure cooperative percep- tion and forecasting,

    H. Yu, W. Yang, H. Ruan, Z. Yang, Y . Tang, X. Gao, X. Hao, Y . Shi, Y . Pan, N. Sun, J. Song, J. Yuan, P. Luo, and Z. Nie, “V2x-seq: A large- scale sequential dataset for vehicle-infrastructure cooperative percep- tion and forecasting,” in Proceedings of the IEEE/CVF Conferen...

    2023

  30. [38]

    Cobevt: Cooperative bird’s eye view semantic segmentation with sparse trans- formers,

    R. Xu, Z. Tu, H. Xiang, W. Shao, B. Zhou, and J. Ma, “Cobevt: Cooperative bird’s eye view semantic segmentation with sparse trans- formers,” arXiv preprint arXiv:2207.02202 , 2022

    2022 arXiv

  31. [39]

    Road- side lidar assisted cooperative localization for connected autonomous vehicles,

    Y . Jiang, E. Javanmard, J. Nakazato, M. Tsukada, and H. Esaki, “Road- side lidar assisted cooperative localization for connected autonomous vehicles,” ACM Intelligent Computing and its Emerging Applications (ICEA), 2023

    2023

  32. [40]

    Lidar-based cooperative relative localization,

    J. Dong, Q. Chen, D. Qu, H. Lu, A. Ganlath, Q. Yang, S. Chen, and S. Labi, “Lidar-based cooperative relative localization,” in 2023 IEEE Intelligent V ehicles Symposium (IV) . IEEE, 2023, pp. 1–8

    2023

  33. [41]

    Learning cooperative trajectory representations for motion forecasting,

    H. Ruan, H. Yu, W. Yang, S. Fan, Y . Tang, and Z. Nie, “Learning cooperative trajectory representations for motion forecasting,” arXiv preprint arXiv:2311.00371, 2023

    2023 arXiv

  34. [42]

    Collaborative semantic occupancy prediction with hybrid feature fusion in connected automated vehicles,

    R. Song, C. Liang, H. Cao, Z. Yan, W. Zimmer, M. Gross, A. Festag, and A. Knoll, “Collaborative semantic occupancy prediction with hybrid feature fusion in connected automated vehicles,” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recog- nition, 2024

    2024

  35. [43]

    V2x-seq: A large-scale sequential dataset for vehicle-infrastructure cooperative perception and forecasting,

    H. Yu, W. Yang, H. Ruan, Z. Yang, Y . Tang, X. Gao, X. Hao, Y . Shi, Y . Pan, N. Sun et al. , “V2x-seq: A large-scale sequential dataset for vehicle-infrastructure cooperative perception and forecasting,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern...

    2023

  36. [44]

    Carla: An open urban driving simulator,

    A. Dosovitskiy, G. Ros, F. Codevilla, A. Lopez, and V . Koltun, “Carla: An open urban driving simulator,” in CoRL, 2017

    2017

  37. [45]

    Airsim: High-fidelity visual and physical simulation for autonomous vehicles,

    S. Shah, D. Dey, C. Lovett, and A. Kapoor, “Airsim: High-fidelity visual and physical simulation for autonomous vehicles,” in Field and Service Robotics: Results of the 11th International Conference . Springer, 2018, pp. 621–635

    2018

  38. [46]

    Lidarsim: Realistic lidar simulation by leveraging the real world,

    S. Manivasagam, S. Wang, K. Wong, W. Zeng, M. Sazanovich, S. Tan, B. Yang, W.-C. Ma, and R. Urtasun, “Lidarsim: Realistic lidar simulation by leveraging the real world,” in CVPR, 2020

    2020

  39. [47]

    Surfelgan: Synthesizing realistic sensor data for autonomous driving,

    Z. Yang, Y . Chai, D. Anguelov, Y . Zhou, P. Sun, D. Erhan, S. Rafferty, and H. Kretzschmar, “Surfelgan: Synthesizing realistic sensor data for autonomous driving,” in CVPR, 2020

    2020

  40. [48]

    Augmented lidar simulator for autonomous driving,

    J. Fang, D. Zhou, F. Yan, T. Zhao, F. Zhang, Y . Ma, L. Wang, and R. Yang, “Augmented lidar simulator for autonomous driving,” IEEE Robotics and Automation Letters , vol. 5, no. 2, pp. 1931–1938, 2020

    1931

  41. [49]

    Reconstructing objects in-the-wild for realistic sensor simulation,

    Z. Yang, S. Manivasagam, Y . Chen, J. Wang, R. Hu, and R. Urtasun, “Reconstructing objects in-the-wild for realistic sensor simulation,” in ICRA, 2023

    2023

  42. [50]

    Geosim: Realistic video simulation via geometry-aware composition for self-driving,

    Y . Chen, F. Rong, S. Duggal, S. Wang, X. Yan, S. Manivasagam, S. Xue, E. Yumer, and R. Urtasun, “Geosim: Realistic video simulation via geometry-aware composition for self-driving,” in CVPR, 2021

    2021

  43. [51]

    Cadsim: Robust and scalable in-the-wild 3d reconstruction for controllable sensor simulation,

    J. Wang, S. Manivasagam, Y . Chen, Z. Yang, I. A. B ˆarsan, A. J. Yang, W.-C. Ma, and R. Urtasun, “Cadsim: Robust and scalable in-the-wild 3d reconstruction for controllable sensor simulation,” in CoRL, 2022

    2022

  44. [52]

    Mega-nerf: Scalable construction of large-scale nerfs for virtual fly-throughs,

    H. Turki, D. Ramanan, and M. Satyanarayanan, “Mega-nerf: Scalable construction of large-scale nerfs for virtual fly-throughs,” in CVPR, 2022

    2022

  45. [53]

    Neural scene graphs for dynamic scenes,

    J. Ost, F. Mannan, N. Thuerey, J. Knodt, and F. Heide, “Neural scene graphs for dynamic scenes,” in CVPR, 2021

    2021

  46. [54]

    Panoptic Neural Fields: A Semantic Object-Aware Neural Scene Representation,

    A. Kundu, K. Genova, X. Yin, A. Fathi, C. Pantofaru, L. Guibas, A. Tagliasacchi, F. Dellaert, and T. Funkhouser, “Panoptic Neural Fields: A Semantic Object-Aware Neural Scene Representation,” in CVPR, 2022

    2022

  47. [55]

    S-nerf: Neural radiance fields for street views,

    Z. Xie, J. Zhang, W. Li, F. Zhang, and L. Zhang, “S-nerf: Neural radiance fields for street views,” in ICLR, 2023

    2023

  48. [56]

    Neurad: Neural rendering for autonomous driving,

    A. Tonderski, C. Lindstr ¨om, G. Hess, W. Ljungbergh, L. Svensson, and C. Petersson, “Neurad: Neural rendering for autonomous driving,” in CVPR, 2024

    2024

  49. [57]

    Periodic vibration gaussian: Dynamic urban scene reconstruction and real-time render- ing,

    Y . Chen, C. Gu, J. Jiang, X. Zhu, and L. Zhang, “Periodic vibration gaussian: Dynamic urban scene reconstruction and real-time render- ing,” arXiv:2311.18561, 2023

    2023 arXiv

  50. [58]

    Who2com: Collaborative perception via learnable handshake com- munication,

    Y .-C. Liu, J. Tian, C.-Y . Ma, N. Glaser, C.-W. Kuo, and Z. Kira, “Who2com: Collaborative perception via learnable handshake com- munication,” in 2020 IEEE International Conference on Robotics and Automation (ICRA) . IEEE, 2020, pp. 6876–6883

    2020

  51. [59]

    When2com: Multi-agent perception via communication graph grouping,

    Y .-C. Liu, J. Tian, N. Glaser, and Z. Kira, “When2com: Multi-agent perception via communication graph grouping,” in Proceedings of the IEEE/CVF Conference on computer vision and pattern recognition , 2020, pp. 4106–4115

    2020

  52. [60]

    Opv2v: An open benchmark dataset and fusion pipeline for perception with vehicle-to- vehicle communication,

    R. Xu, H. Xiang, X. Xia, X. Han, J. Liu, and J. Ma, “Opv2v: An open benchmark dataset and fusion pipeline for perception with vehicle-to- vehicle communication,” 2022 International Conference on Robotics and Automation (ICRA) , pp. 2583–2589, 2021

    2022

  53. [61]

    Learning distilled collaboration graph for multi-agent perception,

    Y . Li, S. Ren, P. Wu, S. Chen, C. Feng, and W. Zhang, “Learning distilled collaboration graph for multi-agent perception,” Advances in Neural Information Processing Systems , vol. 34, pp. 29 541–29 552, 2021

    2021

  54. [62]

    V2x- sim: Multi-agent collaborative perception dataset and benchmark for autonomous driving,

    Y . Li, D. Ma, Z. An, Z. Wang, Y . Zhong, S. Chen, and C. Feng, “V2x- sim: Multi-agent collaborative perception dataset and benchmark for autonomous driving,” IEEE Robotics and Automation Letters , 2022

    2022

  55. [63]

    V2x-vit: Vehicle-to-everything cooperative perception with vision transformer,

    R. Xu, H. Xiang, Z. Tu, X. Xia, M.-H. Yang, and J. Ma, “V2x-vit: Vehicle-to-everything cooperative perception with vision transformer,” in Computer Vision–ECCV 2022: 17th European Conference, Tel Aviv, Israel, October 23–27, 2022, Proceedings, Part XXXIX . Springer, 2022, pp. 107–124

    2022

  56. [64]

    DAIR-V2X: A large-scale dataset for vehicle- infrastructure cooperative 3d object detection,

    H. Yu, Y . Luo, M. Shu, Y . Huo, Z. Yang, Y . Shi, Z. Guo, H. Li, X. Hu, J. Yuan et al. , “DAIR-V2X: A large-scale dataset for vehicle- infrastructure cooperative 3d object detection,” In Proceedings of the IEEE/CVF Conference on computer vision and pattern recognition (CVPR), 2022

    2022

  57. [65]

    CoBEVT: Cooperative bird’s eye view semantic segmentation with sparse trans- formers,

    R. Xu, Z. Tu, H. Xiang, W. Shao, B. Zhou, and J. Ma, “CoBEVT: Cooperative bird’s eye view semantic segmentation with sparse trans- formers,” CoRL, 2022

    2022

  58. [66]

    Where2comm: Communication-efficient collaborative perception via spatial confi- dence maps,

    Y . Hu, S. Fang, Z. Lei, Y . Zhong, and S. Chen, “Where2comm: Communication-efficient collaborative perception via spatial confi- dence maps,” Advances in Neural Information Processing Systems , 2022

    2022

  59. [67]

    Robust collaborative 3d object detection in presence of pose errors,

    Y . Lu, Q. Li, B. Liu, M. Dianat, C. Feng, S. Chen, and Y . Wang, “Robust collaborative 3d object detection in presence of pose errors,” IEEE International Conference on Robotics and Automation (ICRA) , 2023

    2023

  60. [68]

    Multi-robot scene completion: Towards task-agnostic collaborative perception,

    Y . Li, J. Zhang, D. Ma, Y . Wang, and C. Feng, “Multi-robot scene completion: Towards task-agnostic collaborative perception,” in Con- ference on Robot Learning , 2022

    2022

  61. [69]

    Regularized graph matching for correspondence identification under uncertainty in collaborative perception,

    P. Gao, R. Guo, H. Lu, and H. Zhang, “Regularized graph matching for correspondence identification under uncertainty in collaborative perception,” Robotics: Science and Systems XVI , 2020

    2020

  62. [70]

    Pragmatic communication in multi-agent collaborative perception,

    Y . Hu, X. Pang, X. Qin, Y . C. Eldar, S. Chen, P. Zhang, and W. Zhang, “Pragmatic communication in multi-agent collaborative perception,” arXiv preprint arXiv:2401.12694 , 2024

    2024 arXiv

  63. [71]

    Towards collaborative autonomous driving: Simulation platform and end-to-end system,

    G. Liu, Y . Hu, C. Xu, W. Mao, J. Ge, Z. Huang, Y . Lu, Y . Xu, J. Xia, Y . Wanget al., “Towards collaborative autonomous driving: Simulation platform and end-to-end system,” arXiv preprint arXiv:2404.09496 , 2024

    2024 arXiv

  64. [72]

    Neusg: Neural implicit surface reconstruction with 3d gaussian splatting guidance,

    H. Chen, C. Li, and G. H. Lee, “Neusg: Neural implicit surface reconstruction with 3d gaussian splatting guidance,” arXiv preprint arXiv:2312.00846, 2023

    2023 arXiv

  65. [73]

    Gaussian opacity fields: Efficient adaptive surface reconstruction in unbounded scenes,

    Z. Yu, T. Sattler, and A. Geiger, “Gaussian opacity fields: Efficient adaptive surface reconstruction in unbounded scenes,” ACM Transac- tions on Graphics , 2024

    2024

  66. [74]

    Stablenormal: Reducing diffusion variance for stable and sharp normal,

    C. Ye, L. Qiu, X. Gu, Q. Zuo, Y . Wu, Z. Dong, L. Bo, Y . Xiu, and X. Han, “Stablenormal: Reducing diffusion variance for stable and sharp normal,” ACM Transactions on Graphics (TOG) , 2024

    2024

  67. [75]

    Grounded sam: Assembling open-world models for diverse visual tasks,

    T. Ren, S. Liu, A. Zeng, J. Lin, K. Li, H. Cao, J. Chen, X. Huang, Y . Chen, F. Yan, Z. Zeng, H. Zhang, F. Li, J. Yang, H. Li, Q. Jiang, and L. Zhang, “Grounded sam: Assembling open-world models for diverse visual tasks,” 2024

    2024

  68. [76]

    Video k-net: A simple, strong, and unified baseline for video segmentation,

    X. Li, W. Zhang, J. Pang, K. Chen, G. Cheng, Y . Tong, and C. C. Loy, “Video k-net: A simple, strong, and unified baseline for video segmentation,” in CVPR, 2022

    2022

  69. [77]

    Structured 3d latents for scalable and versatile 3d generation,

    J. Xiang, Z. Lv, S. Xu, Y . Deng, R. Wang, B. Zhang, D. Chen, X. Tong, and J. Yang, “Structured 3d latents for scalable and versatile 3d generation,” arXiv preprint arXiv:2412.01506 , 2024

    2024 arXiv

  70. [78]

    Monolss: Learnable sample selection for monocular 3d detection,

    Z. Li, J. Jia, and Y . Shi, “Monolss: Learnable sample selection for monocular 3d detection,” in 2024 International Conference on 3D Vision (3DV). IEEE, 2024, pp. 1125–1135

    2024

  71. [79]

    Bevheight: A robust framework for vision-based roadside 3d object detection,

    L. Yang, K. Yu, T. Tang, J. Li, K. Yuan, L. Wang, X. Zhang, and P. Chen, “Bevheight: A robust framework for vision-based roadside 3d object detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp. 21 611–21 620

    2023

  72. [80]

    Imvoxelnet: Image to voxels projection for monocular and multi-view general-purpose 3d object detection,

    D. Rukhovich, A. V orontsova, and A. Konushin, “Imvoxelnet: Image to voxels projection for monocular and multi-view general-purpose 3d object detection,” in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , 2022, pp. 2397–2406

    2022

  73. [81]

    Hugs: Holistic urban 3d scene understanding via gaussian splatting,

    H. Zhou, J. Shao, L. Xu, D. Bai, W. Qiu, B. Liu, Y . Wang, A. Geiger, and Y . Liao, “Hugs: Holistic urban 3d scene understanding via gaussian splatting,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 21 336–21 345

    2024

  74. [82]

    Occ3d: A large-scale 3d occupancy prediction benchmark for autonomous driving,

    X. Tian, T. Jiang, L. Yun, Y . Mao, H. Yang, Y . Wang, Y . Wang, and H. Zhao, “Occ3d: A large-scale 3d occupancy prediction benchmark for autonomous driving,” Advances in Neural Information Processing Systems, vol. 36, pp. 64 318–64 330, 2023

    2023

  75. [83]

    Deraings: Gaussian splatting for enhanced scene reconstruction in rainy,

    S. Liu, X. Chen, H. Chen, Q. Xu, and M. Li, “Deraings: Gaussian splatting for enhanced scene reconstruction in rainy,” arXiv e-prints , pp. arXiv–2408, 2024

    2024

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.