Pith. sign in

REVIEW 2 major objections 6 minor 1 cited by

SafeMap: Robust HD Map Construction from Incomplete Observations

T0 review · 2 major / 6 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 Two plug-and-play modules keep HD maps accurate when cameras fail

desk verdict Useful first application of masked-view reconstruction to HD map construction, but the reported gains are confounded with the random-mask training schedule and need a proper control baseline before the modules get the credit. read the letter →

arxiv 2507.00861 v1 pith:V7E655E5 submitted 2025-07-01 cs.CV

Xiaoshuai Hao , Lingdong Kong , Rong Yin , Pengwei Wang , Jing Zhang , Yunfeng Diao , Shu Zhao This is my paper
classification cs.CV
keywords HDmapconstructionincompletemulti-viewobservationscamerarobustnessGaussianreferencepointsBEVfeaturedistillationautonomousdrivingnuScenesArgoverse2
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

The paper tries to establish that HD map construction, which normally assumes all surround-view cameras are working, can be made robust to missing camera views by two lightweight plug-and-play modules. The first module, G-PVR, reconstructs a missing perspective view by sampling Gaussian-distributed reference points across a panorama stitched from the available views and letting deformable attention pull information from the views most relevant to the gap. The second, D-BEVC, corrects the bird's-eye-view features computed from incomplete observations by distilling them toward the full-view BEV features with an MSE loss. On nuScenes, the method reports mAP gains of 2.4 to 18.2 points over the HIMap baseline across six single-missing-view cases, with the largest gains when the front or back camera is lost. If these results hold, existing HD map models can gain missing-camera robustness without an architectural redesign.

What carries the argument

The load-bearing mechanism is the Gaussian-based reference-point sampler in G-PVR. The available perspective features are concatenated into one panoramic feature map, and the missing view's query samples that map at points whose horizontal coordinate is drawn from a Gaussian centered on the missing view's location in the panorama; the variance $\sigma^2$ controls how far sampling reaches. Because deformable attention then aggregates keys and values at those reference points, the reconstruction is spatially biased toward adjacent views while still able to reach opposite views. D-BEVC supplies a second mechanism: an MSE loss between complete BEV features and BEV features derived from masked views, so the BEV representation itself is pulled toward the full-observation feature space during training.

What would settle it

Evaluate SafeMap on test cases where two or three adjacent cameras drop out together, or where the same physical camera is missing for an entire sequence, and compare mAP against the random single-view masking numbers in Tables 1 and 2; if mAP falls back to baseline levels, the robustness claim is conditional on the training mask distribution.

Watch

Extended reading notes

Core claim

The central claim is that missing-view robustness for online vectorized HD map construction can be obtained without a new detection or map framework: the paper's two modules can be dropped onto MapTR or HIMap and retrained with random view masking. G-PVR treats the missing view as a learnable query over a panorama of all available views, with reference points drawn from a Gaussian centered on the position of the missing view in the stitched panorama, so adjacent views contribute more than distant ones. D-BEVC adds a distillation loss that pulls BEV features from incomplete observations toward the complete-view BEV features. With both modules, the paper reports consistent gains in every missing-view scenario on nuScenes and Argoverse2, including a 16.3-point mAP recovery when the front view is missing on HIMap and an 18.2-point recovery when the back view is missing.

Load-bearing premise

Randomly dropping one camera view during training and testing faithfully represents real-world incomplete inputs; if real failures are correlated or multi-view, the reported robustness may not hold.

Editorial extensions

If this is right

  • On nuScenes, adding SafeMap to HIMap raises mAP from 38.0 to 54.3 when the front camera is missing and from 33.1 to 51.3 when the back camera is missing.
  • The method also improves complete-view performance slightly (MapTR 50.3 to 52.5; HIMap 65.5 to 66.0), so the reconstruction modules do not trade robustness for normal-case accuracy.
  • In the averaged missing-view setting, G-PVR alone and D-BEVC alone each add about 6.6 mAP points over the MapTR baseline, and together they add 7.4 points.
  • With up to five of six views missing, SafeMap keeps a margin over MapTR: 6.0 versus 4.0 mAP for five missing views.
  • The modules add only 0.4M to 3.6M parameters and no meaningful speed or memory cost, so they are compatible with existing pipelines.

Reading between the lines

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

  • Inference: because the largest gains appear when the front or back camera is missing, HD map systems could use SafeMap's view-importance prior to decide which camera to protect or repair first; the paper does not draw this operational conclusion.
  • Inference: the same Gaussian-distance weighting could be applied beyond maps, for instance to BEV segmentation or 3D detection with missing views, since G-PVR is architecture-agnostic; this is untested in the paper.
  • Inference: a real deployment test with physically corrupted or occluded lenses, rather than masked pixels, would show whether the reconstruction prior survives photometric corruption such as lens dirt, motion blur, or glare, which the paper's random masking does not model.
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

2 major / 6 minor

Summary. The paper proposes SafeMap, a plug-and-play extension to camera-based HD map construction for settings where some camera views are missing. During training, views are randomly masked (Random View Masking, RVM). SafeMap adds two modules to a base map-construction model such as MapTR or HIMap: a Gaussian-based Perspective View Reconstruction (G-PVR) module that reconstructs the missing perspective-view features from available views using deformable attention with Gaussian-sampled reference points, and a Distillation-based BEV Correction (D-BEVC) module that trains the incomplete-observation BEV features to match BEV features produced from complete observations. Experiments on nuScenes and Argoverse2 report improvements over MapTR and HIMap in both complete and incomplete observation settings, together with component ablations, sensitivity analyses, and robustness evaluations under sensor corruptions.

Significance. If the reported gains are truly attributable to the proposed G-PVR and D-BEVC modules, SafeMap would be a useful, low-overhead robustness technique for HD map construction, a problem that is indeed less explored than robustness for 3D object detection. The paper has concrete strengths: per-view results are reported for six and seven camera configurations, the ablations are systematic, the added parameter and speed costs are small, and the tables are internally consistent. However, the central attribution is currently not isolated. As detailed below, the ablation baseline appears to be the original MapTR trained without the random view masking schedule, so the headline gains conflate the proposed modules with the RVM training distribution itself. No code or formal proofs are provided, so the contribution rests on experimental isolation, and that isolation is missing. With a properly controlled baseline, the claim could be credible; without it, the abstract's statement that G-PVR and D-BEVC enable robust reconstruction is not yet supported.

major comments (2)
  1. [Section 4.3, Table 4; compare Table 1] The baseline row in Table 4 reports an average missing-view mAP of 39.9 (APped 36.4, APdiv 42.0, APbou 41.5), which is exactly the average of the MapTR missing-view rows in Table 1 (39.9, 36.4, 42.0, 41.5). The text in Section 4.1 says SafeMap is trained by randomly discarding RGB images, while the baseline models are retrained according to their default configurations. This strongly indicates that the Table 4 baseline is MapTR trained without the Random View Masking schedule. Consequently, the 7.4-point gap between this baseline and full SafeMap conflates two changes: (i) adding G-PVR and D-BEVC and their losses, and (ii) training on randomly masked views at all. The paper never reports a control trained with RVM but without G-PVR and D-BEVC. Please add such a control to Table 4, and also retrain MapTR and HIMap with the same RVM schedule without the proposed modules for the Section 4.2 comparisons; otherwise the claimed benefits of the modules over the baselines are not identifiable.
  2. [Section 4.2, Tables 2 and 3] The system-level comparisons against HIMap (Table 2) and MapTR on Argoverse2 (Table 3) share the same confound. The baselines are described as retrained 'according to their default settings,' i.e., without random view masking, whereas all SafeMap variants are trained with RVM. The reported gains in the incomplete-view rows (e.g., 2.4 to 18.2 percentage points over HIMap) therefore do not isolate the contribution of G-PVR and D-BEVC from the contribution of simply seeing masked-view patterns during training. A masked-training-only baseline is required before the abstract's claim that SafeMap 'significantly outperforms previous methods' in incomplete scenarios can be attributed to the proposed architecture.
minor comments (6)
  1. [Section 3.3, Eq. (3)] Equation (3) uses F_incom_PV on both sides: first as the incomplete input feature and then, immediately after the equation, as the decoder output (F_incom_PV = Decoder([V, F_PPV])). Please use a distinct symbol such as F_rec_PV for the reconstructed feature to avoid this notational collision.
  2. [Introduction and Conclusion] The module is called G-GVR (Gaussian-based Global View Reconstruction) in the Introduction and Conclusion, but G-PVR (Gaussian-based Perspective View Reconstruction) in the contributions, methodology, and experiments. Standardize the name throughout.
  3. [Section 4.2] The sentence 'SafeMap improves the HIMap model's mAP metric by 2.4% to 18.2%' should be phrased as 'by 2.4 to 18.2 percentage points' to avoid ambiguity with relative percentages; the table values are absolute mAP differences.
  4. [Section 4.2 and Table 8] The text says SafeMap increases the number of parameters 'by 0.4MB to 3.6MB'; the table reports parameters in millions, so this should be '0.4M to 3.6M parameters'.
  5. [Section 4.4] The phrase 'miss all view setting' is unclear; the sensitivity analysis appears to concern missing one view (the front view, based on Fig. 5) rather than all views. Please rephrase to describe the exact experimental setting.
  6. [Figure 2 and Section 3.2] The acronym RVM appears in the Figure 2 caption but is not defined in Section 3.2 where the masking procedure is first described. Define it at first use.

Circularity Check

1 steps flagged · score 6.0 of 10

Module ablation baseline is untrained with RVM, so the claimed G-PVR/D-BEVC gain reduces to a training-protocol confound.

  1. other [Section 4.3, Table 4; Section 4.1 Implementation Details; Table 1]
    "The following ablation models were designed: 1) SafeMap (Baseline): the model trained without the reconstruction module; ... ✗ ✗ 36.4 42.0 41.5 39.9"

    The Table 4 'Baseline' row (39.9 mAP) is exactly the average of the Table 1 MapTR missing-view rows (31.3, 43.7, 44.2, 28.5, 45.8, 46.1). Section 4.1 says baselines were 'retrained using their configurations' and only SafeMap is trained by 'randomly discard RGB images of any view'. Thus the baseline was not trained with the Random View Masking (RVM) scheme, while the SafeMap rows were. The 7.4-point gain attributed to G-PVR and D-BEVC therefore conflates the new modules with the RVM training protocol; the paper provides no RVM-only control. The conclusion that 'each module in SafeMap meaningfully contributes' is forced by the construction of the control condition, not by an isolated module effect.

full rationale

The one load-bearing circular/confounded step is the Table 4 ablation: the baseline is numerically identical to Table 1's MapTR missing-view average (39.9 mAP), which per Section 4.1 was retrained with the original configurations, i.e., without the Random View Masking scheme used for SafeMap. Consequently the 7.4-point improvement credited to G-PVR+D-BEVC is not an isolated module effect; it conflates the new modules with the RVM training distribution. No other circularity was found: the map head is supervised by external ground-truth maps (L_map in Eq. 5), and the reconstruction/distillation losses (Eqs. 3-4) are self-supervised auxiliary objectives whose final evaluation does not reduce to their own targets. Self-citations to MapBench are used only as an evaluation protocol, not as load-bearing support. The Gaussian reference-point prior is an explicit design assumption, not a derived prediction.

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

SafeMap introduces two new modules but no new physical entities, particles, or conserved quantities. The central claim depends on three free hyperparameters (λ1, λ2, σ) and several domain assumptions, most notably that random masking simulates real failures and that the Gaussian prior puts attention where it is needed.

free parameters (3)
  • λ1 = 0.05
    Weight for the perspective view reconstruction loss, tuned on nuScenes validation (Fig. 5).
  • λ2 = 5
    Weight for the BEV feature correction loss, tuned on nuScenes validation (Fig. 5).
  • σ = 3
    Variance of the Gaussian used to sample reference points in Eq. 2, tuned on nuScenes validation (Fig. 6).
assumptions (5)
  • domain assumption Random view masking during training is a valid simulation of real camera failures.
    The method and evaluation rely on this to generalize to test-time missing views (Section 3.2, Section 4.1).
  • domain assumption Available camera views contain enough information to reconstruct the missing view.
    G-PVR assumes the remaining views provide the cues needed for reconstruction (Section 3.3).
  • ad hoc to paper A Gaussian centered at the panorama midpoint concentrates attention on the most informative regions.
    This prior is not derived or validated independently; it is chosen to favor the views adjacent to the missing one (Eq. 2).
  • domain assumption Complete BEV features from the same model are an appropriate supervisory target for incomplete BEV features.
    D-BEVC uses the model's own full-view output as the correction target (Section 3.4).
  • domain assumption The nuScenes and Argoverse2 validation sets are representative of the map-construction task.
    All conclusions are drawn from these standard benchmarks (Section 4.1).

how reviews work

0 comments
Cite this review

Pith. "Pith review of SafeMap: Robust HD Map Construction from Incomplete Observations." pith.science (2026). https://pith.science/paper/V7E655E5

@misc{pith2026250700861,
  author       = {Pith},
  title        = {Pith review of: SafeMap: Robust HD Map Construction from Incomplete Observations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/V7E655E5}},
  note         = {Machine review of arXiv:2507.00861}
}
read the original abstract

Robust high-definition (HD) map construction is vital for autonomous driving, yet existing methods often struggle with incomplete multi-view camera data. This paper presents SafeMap, a novel framework specifically designed to secure accuracy even when certain camera views are missing. SafeMap integrates two key components: the Gaussian-based Perspective View Reconstruction (G-PVR) module and the Distillation-based Bird's-Eye-View (BEV) Correction (D-BEVC) module. G-PVR leverages prior knowledge of view importance to dynamically prioritize the most informative regions based on the relationships among available camera views. Furthermore, D-BEVC utilizes panoramic BEV features to correct the BEV representations derived from incomplete observations. Together, these components facilitate the end-to-end map reconstruction and robust HD map generation. SafeMap is easy to implement and integrates seamlessly into existing systems, offering a plug-and-play solution for enhanced robustness. Experimental results demonstrate that SafeMap significantly outperforms previous methods in both complete and incomplete scenarios, highlighting its superior performance and reliability.

Figures

Figures reproduced from arXiv: 2507.00861 by the authors.

Figure 1
Figure 1. Online Vectorized HD Map Construction with Incom￾plete Observations. C/IC denotes complete/incomplete observa￾tions. IC-F/FL/FR/B/BL/BR represents a collection of six scenar￾ios of missing camera views, where F, B, L, and R correspond to front, back, left, and right camera views, respectively. cise and detailed static environmental information essential for vehicle planning and navigation (Caesar et al., 2020; Zhang… view at source ↗
Figure 2
Figure 2. Overview of the SafeMap Framework. We first extract features from complete multi-view camera images and efficiently transform them into a unified BEV space using view transformations. To simulate emergency scenarios involving camera failures, we employ a Random View Masking (RVM) and recovery scheme. Specifically, we introduce a novel Gaussian-based Perspective View Reconstruction (G-PVR) module and a Distillation-b… view at source ↗
Figure 3
Figure 3. Illustration of the proposed Gaussian-based Perspective View Reconstruction (G-PVR) module. the panoramic perspective view: px ∼ N (Na ∗ W/2; σ 2 ), py ∼ U(0, H), (2) where N and U is the Gaussian distribution and uniform distribution, respectively. σ 2 is the variance. G-PVR module effectively guides the module in focusing on the most informative regions of the input views, and can easily adapt to different numbers… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Comparisons of the Gaussian-based PVR vs. Local PVR. ft-N denotes fine-tuning N epochs. G-PVR): the model trained with the Gaussian-based Per￾spective View Reconstruction (G-PVR) module; 3) SafeMap (w/ D-BEVC): the model trained with the Distill-based BEV Correction (D…
Figure 5
Figure 5. Figure 5: Sensitivity Analysis of Hyperparameters λ1 and λ2. formance against eight real-world camera sensor corruptions from MapBench (Hao et al., 2024a), categorized into exte￾rior environments, interior sensors, and sensor failures. The details of these corruptions are outlin…
Figure 6
Figure 6. Figure 6: Sensitivity Analysis of Hyperparameter Variance σ. 4.4. Sensitivity of Hyper-parameters Sensitivity Analysis of Hyperparameters λ1 and λ2. We analyze the sensitivity of hyperparameters λ1 and λ2, as illustrated in [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Qualitative Comparisons. The camera view marked with the symbol indicates the absence of this perspective. shown, progressively increasing λ1 from 0.01 to 0.09 leads to consistent improvements, achieving optimal performance at λ1 = 0.05. Additionally, [PITH_FULL_IMAGE…

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

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

  1. DSBench: A Comprehensive Benchmark for Evaluating External and In-Cabin Risks

    cs.RO 2025-11 reject novelty 5.0 of 10

    A new benchmark claims to be the first to test VLMs on both external and in-cabin driving risks, and reports a fine-tuned model far outperforming all baselines.

Reference graph

Works this paper leans on

64 extracted references · 54 canonical work pages · cited by 1 Pith paper

  1. [1]

    H., Vora, S., Liong, V

    Caesar, H., Bankiti, V., Lang, A. H., Vora, S., Liong, V. E., et al. nuScenes : A multimodal dataset for autonomous driving. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 11618--11628, 2020

    work page 2020

  2. [2]

    End-to-end autonomous driving: Challenges and frontiers

    Chen, L., Wu, P., Chitta, K., Jaeger, B., Geiger, A., and Li, H. End-to-end autonomous driving: Challenges and frontiers. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024 a

    work page 2024

  3. [3]

    Efficient and robust 2d-to-bev representation learning via geometry-guided kernel transformer

    Chen, S., Cheng, T., Wang, X., Meng, W., Zhang, Q., and Liu, W. Efficient and robust 2d-to-bev representation learning via geometry-guided kernel transformer. arXiv preprint arXiv:2206.04584, 2022

    arXiv 2022

  4. [4]

    M-bev: Masked bev perception for robust autonomous driving

    Chen, S., Ma, Y., Qiao, Y., and Wang, Y. M-bev: Masked bev perception for robust autonomous driving. In AAAI Conference on Artificial Intelligence, volume 38, pp.\ 1183--1191, 2024 b

    work page 2024

  5. [5]

    Stvit+: improving self-supervised multi-camera depth estimation with spatial-temporal context and adversarial geometry regularization

    Chen, Z., Zhao, H., Hao, X., Yuan, B., and Li, X. Stvit+: improving self-supervised multi-camera depth estimation with spatial-temporal context and adversarial geometry regularization. Applied Intelligence, 55 0 (5): 0 328, 2025

    work page 2025

  6. [6]

    Pivotnet: Vectorized pivot learning for end-to-end hd map construction

    Ding, W., Qiao, L., Qiu, X., and Zhang, C. Pivotnet: Vectorized pivot learning for end-to-end hd map construction. In IEEE/CVF International Conference on Computer Vision, pp.\ 3672--3682, 2023

    work page 2023

  7. [7]

    Superfusion: Multilevel lidar-camera fusion for long-range hd map generation

    Dong, H., Gu, W., Zhang, X., Xu, J., Ai, R., Lu, H., Kannala, J., and Chen, X. Superfusion: Multilevel lidar-camera fusion for long-range hd map generation. In IEEE International Conference on Robotics and Automation, pp.\ 9056--9062, 2024

    work page 2024

  8. [8]

    Benchmarking robustness of 3d object detection to common corruptions

    Dong, Y., Kang, C., Zhang, J., Zhu, Z., Wang, Y., Yang, X., Su, H., Wei, X., and Zhu, J. Benchmarking robustness of 3d object detection to common corruptions. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 1022--1032, 2023

    work page 2023

Show all 64 references
  1. [9]

    Metabev: Solving sensor failures for 3d detection and map segmentation

    Ge, C., Chen, J., Xie, E., Wang, Z., Hong, L., Lu, H., Li, Z., and Luo, P. Metabev: Solving sensor failures for 3d detection and map segmentation. In IEEE/CVF International Conference on Computer Vision, pp.\ 8721--8731, 2023

    2023

  2. [10]

    Is your hd map constructor reliable under sensor corruptions? In Advances in Neural Information Processing Systems, volume 37, 2024 a

    Hao, X., Wei, M., Yang, Y., Zhao, H., Zhang, H., Zhou, Y., Wang, Q., Li, W., Kong, L., and Zhang, J. Is your hd map constructor reliable under sensor corruptions? In Advances in Neural Information Processing Systems, volume 37, 2024 a

    2024

  3. [11]

    Mbfusion: A new multi-modal bev feature fusion method for hd map construction

    Hao, X., Zhang, H., Yang, Y., Zhou, Y., Jung, S., Park, S.-I., and Yoo, B. Mbfusion: A new multi-modal bev feature fusion method for hd map construction. In IEEE International Conference on Robotics and Automation, pp.\ 15922--15928, 2024 b

    2024

  4. [12]

    Mapfusion: A novel bev feature fusion network for multi-modal map construction

    Hao, X., Diao, Y., Wei, M., Yang, Y., Hao, P., Yin, R., Zhang, H., Li, W., Zhao, S., and Liu, Y. Mapfusion: A novel bev feature fusion network for multi-modal map construction. Information Fusion, 119: 0 103018, 2025 a

    2025

  5. [13]

    Mapdistill: Boosting efficient camera-based hd map construction via camera-lidar fusion model distillation

    Hao, X., Li, R., Zhang, H., Li, D., Yin, R., Jung, S., Park, S.-I., Yoo, B., Zhao, H., and Zhang, J. Mapdistill: Boosting efficient camera-based hd map construction via camera-lidar fusion model distillation. In European Conference on Computer Vision, pp.\ 166--183, 2025 b

    2025

  6. [14]

    Msc-bench: Benchmarking and analyzing multi-sensor corruption for driving perception

    Hao, X., Liu, G., Zhao, Y., Ji, Y., Wei, M., Zhao, H., Kong, L., Yin, R., and Liu, Y. Msc-bench: Benchmarking and analyzing multi-sensor corruption for driving perception. arXiv preprint arXiv:2501.01037, 2025 c

    2025 arXiv

  7. [15]

    Deep residual learning for image recognition

    He, K., Zhang, X., Ren, S., and Sun, J. Deep residual learning for image recognition. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2016

    2016

  8. [16]

    Unified 3d and 4d panoptic segmentation via dynamic shifting networks

    Hong, F., Kong, L., Zhou, H., Zhu, X., Li, H., and Liu, Z. Unified 3d and 4d panoptic segmentation via dynamic shifting networks. IEEE Transactions on Pattern Analysis and Machine Intelligence, 46 0 (5): 0 3480--3495, 2024

    2024

  9. [17]

    C., and Schindler, K

    Ke, B., Obukhov, A., Huang, S., Metzger, N., Daudt, R. C., and Schindler, K. Repurposing diffusion-based image generators for monocular depth estimation. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 9492--9502, 2024

    2024

  10. [18]

    Rethinking range view representation for lidar segmentation

    Kong, L., Liu, Y., Chen, R., Ma, Y., Zhu, X., Li, Y., Hou, Y., Qiao, Y., and Liu, Z. Rethinking range view representation for lidar segmentation. In IEEE/CVF International Conference on Computer Vision, pp.\ 228--240, 2023 a

    2023

  11. [19]

    Robo3d: Towards robust and reliable 3d perception against corruptions

    Kong, L., Liu, Y., Li, X., Chen, R., Zhang, W., Ren, J., Pan, L., Chen, K., and Liu, Z. Robo3d: Towards robust and reliable 3d perception against corruptions. In IEEE/CVF International Conference on Computer Vision, pp.\ 19994--20006, 2023 b

    2023

  12. [20]

    X., Cottereau, B

    Kong, L., Xie, S., Hu, H., Ng, L. X., Cottereau, B. R., and Ooi, W. T. Robodepth: Robust out-of-distribution depth estimation under corruptions. In Adv. Neural Inf. Process. Syst., volume 36, 2023 c

    2023

  13. [21]

    The robodrive challenge: Drive anytime anywhere in any condition

    Kong, L., Xie, S., Hu, H., Niu, Y., et al. The robodrive challenge: Drive anytime anywhere in any condition. arXiv preprint arXiv:2405.08816, 2024

    2024 arXiv

  14. [22]

    T., and Liu, Z

    Kong, L., Xu, X., Ren, J., Zhang, W., Pan, L., Chen, K., Ooi, W. T., and Liu, Z. Multi-modal data-efficient 3d scene understanding for autonomous driving. IEEE Transactions on Pattern Analysis and Machine Intelligence, 47 0 (5): 0 3748--3765, 2025

    2025

  15. [23]

    Hdmapnet: An online hd map construction and evaluation framework

    Li, Q., Wang, Y., Wang, Y., and Zhao, H. Hdmapnet: An online hd map construction and evaluation framework. In IEEE International Conference on Robotics and Automation, pp.\ 4628--4634, 2022 a

    2022

  16. [24]

    Mapnext: Revisiting training and scaling practices for online vectorized hd map construction

    Li, T. Mapnext: Revisiting training and scaling practices for online vectorized hd map construction. arXiv preprint arXiv:2401.07323, 2024

    2024 arXiv

  17. [25]

    Is your lidar placement optimized for 3d scene understanding? In Advances in Neural Information Processing Systems, volume 37, pp.\ 34980--35017, 2024

    Li, Y., Kong, L., Hu, H., Xu, X., and Huang, X. Is your lidar placement optimized for 3d scene understanding? In Advances in Neural Information Processing Systems, volume 37, pp.\ 34980--35017, 2024

    2024

  18. [26]

    Bevformer: Learning bird's-eye-view representation from multi-camera images via spatiotemporal transformers

    Li, Z., Wang, W., Li, H., Xie, E., Sima, C., Lu, T., Qiao, Y., and Dai, J. Bevformer: Learning bird's-eye-view representation from multi-camera images via spatiotemporal transformers. In ECCV, pp.\ 1--18, 2022 b

    2022

  19. [27]

    Maptr: Structured modeling and learning for online vectorized hd map construction

    Liao, B., Chen, S., Wang, X., Cheng, T., Zhang, Q., Liu, W., and Huang, C. Maptr: Structured modeling and learning for online vectorized hd map construction. In International Conference on Learning Representations, 2023 a

    2023

  20. [28]

    Maptrv2: An end-to-end framework for online vectorized HD map construction

    Liao, B., Chen, S., Zhang, Y., Jiang, B., Zhang, Q., Liu, W., Huang, C., and Wang, X. Maptrv2: An end-to-end framework for online vectorized HD map construction. arXiv preprint arXiv:2308.05736, 2023 b

    2023 arXiv

  21. [29]

    Mgmap: Mask-guided learning for online vectorized hd map construction

    Liu, X., Wang, S., Li, W., Yang, R., Chen, J., and Zhu, J. Mgmap: Mask-guided learning for online vectorized hd map construction. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 14812--14821, 2024 a

    2024

  22. [30]

    Segment any point cloud sequences by distilling vision foundation models

    Liu, Y., Kong, L., Cen, J., Chen, R., Zhang, W., Pan, L., Chen, K., and Liu, Z. Segment any point cloud sequences by distilling vision foundation models. In Advances in Neural Information Processing Systems, volume 36, pp.\ 37193--37229, 2023 a

    2023

  23. [31]

    Vectormapnet: End-to-end vectorized hd map learning

    Liu, Y., Yuan, T., Wang, Y., Wang, Y., and Zhao, H. Vectormapnet: End-to-end vectorized hd map learning. In International Conference on Machine Learning, pp.\ 22352--22369, 2023 b

    2023

  24. [32]

    Multi-space alignments towards universal lidar segmentation

    Liu, Y., Kong, L., Wu, X., Chen, R., Li, X., Pan, L., Liu, Z., and Ma, Y. Multi-space alignments towards universal lidar segmentation. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 14648--14661, 2024 b

    2024

  25. [33]

    Swin transformer: Hierarchical vision transformer using shifted windows

    Liu, Z., Lin, Y., Cao, Y., Hu, H., Wei, Y., Zhang, Z., Lin, S., and Guo, B. Swin transformer: Hierarchical vision transformer using shifted windows. In IEEE/CVF International Conference on Computer Vision, pp.\ 9992--10002, 2021

    2021

  26. [34]

    L., and Han, S

    Liu, Z., Tang, H., Amini, A., Yang, X., Mao, H., Rus, D. L., and Han, S. Bevfusion: Multi-task multi-sensor fusion with unified bird's eye view representation. In IEEE International Conference on Robotics and Automation, pp.\ 2774--2781, 2023 c

    2023

  27. [35]

    and Hutter, F

    Loshchilov, I. and Hutter, F. Decoupled weight decay regularization. In International Conference on Learning Representations, 2019

    2019

  28. [36]

    Vision-centric bev perception: A survey

    Ma, Y., Wang, T., Bai, X., Yang, H., Hou, Y., Wang, Y., Qiao, Y., Yang, R., Manocha, D., and Zhu, X. Vision-centric bev perception: A survey. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024

    2024

  29. [37]

    Analysis of euclidean distance and manhattan distance measure in face recognition

    Malkauthekar, M. Analysis of euclidean distance and manhattan distance measure in face recognition. In International Conference on Computational Intelligence and Information Technology, pp.\ 503--507, 2013

    2013

  30. [38]

    Calibrating deep neural networks using focal loss

    Mukhoti, J., Kulharia, V., Sanyal, A., Golodetz, S., Torr, P., and Dokania, P. Calibrating deep neural networks using focal loss. Advances in Neural Information Processing Systems, pp.\ 15288--15299, 2020

    2020

  31. [39]

    End-to-end vectorized hd-map construction with piecewise bezier curve

    Qiao, L., Ding, W., Qiu, X., and Zhang, C. End-to-end vectorized hd-map construction with piecewise bezier curve. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 13218--13228, 2023

    2023

  32. [40]

    Multi-lidar localization and mapping pipeline for urban autonomous driving

    Sauerbeck, F., Kulmer, D., Pielmeier, M., Leitenstern, M., Wei , C., and Betz, J. Multi-lidar localization and mapping pipeline for urban autonomous driving. In 2023 IEEE SENSORS, pp.\ 1--4, 2023

    2023

  33. [41]

    Robustness-aware 3d object detection in autonomous driving: A review and outlook

    Song, Z., Liu, L., Jia, F., Luo, Y., Jia, C., Zhang, G., Yang, L., and Wang, L. Robustness-aware 3d object detection in autonomous driving: A review and outlook. IEEE Transactions on Intelligent Transportation Systems, 2024

    2024

  34. [42]

    Graphbev: Towards robust bev feature alignment for multi-modal 3d object detection

    Song, Z., Yang, L., Xu, S., Liu, L., Xu, D., Jia, C., Jia, F., and Wang, L. Graphbev: Towards robust bev feature alignment for multi-modal 3d object detection. In European Conference on Computer Vision, pp.\ 347--366. Springer, 2025

    2025

  35. [43]

    Sparseocc: Rethinking sparse latent representation for vision-based semantic occupancy prediction

    Tang, P., Wang, Z., Wang, G., Zheng, J., Ren, X., Feng, B., and Ma, C. Sparseocc: Rethinking sparse latent representation for vision-based semantic occupancy prediction. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 15035--15044, 2024

    2024

  36. [44]

    Wang, S., Caesar, H., Nan, L., and Kooij, J. F. Unibev: Multi-modal 3d object detection with uniform bev encoders for robustness against missing sensor modalities. In IEEE Intelligent Vehicles Symposium, pp.\ 2776--2783. IEEE, 2024

    2024

  37. [45]

    Surrounddepth: Entangling surrounding views for self-supervised multi-camera depth estimation

    Wei, Y., Zhao, L., Zheng, W., Zhu, Z., Rao, Y., Huang, G., Lu, J., and Zhou, J. Surrounddepth: Entangling surrounding views for self-supervised multi-camera depth estimation. In Conference on Robot Learning, pp.\ 539--549, 2023 a

    2023

  38. [46]

    Surroundocc: Multi-camera 3d occupancy prediction for autonomous driving

    Wei, Y., Zhao, L., Zheng, W., Zhu, Z., Zhou, J., and Lu, J. Surroundocc: Multi-camera 3d occupancy prediction for autonomous driving. In IEEE/CVF International Conference on Computer Vision, pp.\ 21729--21740, 2023 b

    2023

  39. [47]

    Argoverse 2: Next generation datasets for self-driving perception and forecasting

    Wilson, B., Qi, W., Agarwal, T., et al. Argoverse 2: Next generation datasets for self-driving perception and forecasting. In Advances in Neural Information Processing Systems, 2021

    2021

  40. [48]

    I., Zhang, H., Park, S., and Jung, S

    Xiaoshuai, H., Zhou, Y., Yoo, B. I., Zhang, H., Park, S., and Jung, S. Electronic device and method with birds-eye-view image processing, March 13 2025. US Patent App. 18/883,424

    2025

  41. [49]

    Robobev: Towards robust bird's eye view perception under corruptions

    Xie, S., Kong, L., Zhang, W., Ren, J., Pan, L., Chen, K., and Liu, Z. Robobev: Towards robust bird's eye view perception under corruptions. arXiv preprint arXiv:2304.06719, 2023

    2023 arXiv

  42. [50]

    Benchmarking and improving bird’s eye view perception robustness in autonomous driving

    Xie, S., Kong, L., Zhang, W., Ren, J., Pan, L., Chen, K., and Liu, Z. Benchmarking and improving bird’s eye view perception robustness in autonomous driving. IEEE Transactions on Pattern Analysis and Machine Intelligence, 47 0 (5): 0 3878--3894, 2025

    2025

  43. [51]

    4d contrastive superflows are dense 3d representation learners

    Xu, X., Kong, L., Shuai, H., Zhang, W., Pan, L., Chen, K., Liu, Z., and Liu, Q. 4d contrastive superflows are dense 3d representation learners. In European Conference on Computer Vision, pp.\ 58--80, 2024

    2024

  44. [52]

    Frnet: Frustum-range networks for scalable lidar segmentation

    Xu, X., Kong, L., Shuai, H., and Liu, Q. Frnet: Frustum-range networks for scalable lidar segmentation. IEEE Transactions on Image Processing, 34: 0 2173--2186, 2025

    2025

  45. [53]

    Is-fusion: Instance-scene collaborative fusion for multimodal 3d object detection

    Yin, J., Shen, J., Chen, R., Li, W., Yang, R., Frossard, P., and Wang, W. Is-fusion: Instance-scene collaborative fusion for multimodal 3d object detection. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 14905--14915, 2024

    2024

  46. [54]

    Streammapnet: Streaming mapping network for vectorized online hd map construction

    Yuan, T., Liu, Y., Wang, Y., Wang, Y., and Zhao, H. Streammapnet: Streaming mapping network for vectorized online hd map construction. In IEEE/CVF Winter Conference on Applications of Computer Vision, pp.\ 7356--7365, 2024

    2024

  47. [55]

    Online map vectorization for autonomous driving: A rasterization perspective

    Zhang, G., Lin, J., Wu, S., Song, Y., Luo, Z., Xue, Y., Lu, S., and Wang, Z. Online map vectorization for autonomous driving: A rasterization perspective. In Advances in Neural Information Processing Systems, 2023

    2023

  48. [56]

    Mapnav: A novel memory representation via annotated semantic maps for vlm-based vision-and-language navigation

    Zhang, L., Hao, X., Xu, Q., Zhang, Q., Zhang, X., Wang, P., Zhang, J., Wang, Z., Zhang, S., and Xu, R. Mapnav: A novel memory representation via annotated semantic maps for vlm-based vision-and-language navigation. arXiv preprint arXiv:2502.13451, 2025

    2025 arXiv

  49. [57]

    Enhancing vectorized map perception with historical rasterized maps

    Zhang, X., Liu, G., Liu, Z., Xu, N., Liu, Y., and Zhao, J. Enhancing vectorized map perception with historical rasterized maps. arXiv preprint arXiv:2409.00620, 2024

    2024 arXiv

  50. [58]

    Beverse: Unified perception and prediction in birds-eye-view for vision-centric autonomous driving

    Zhang, Y., Zhu, Z., Zheng, W., Huang, J., Huang, G., Zhou, J., and Lu, J. Beverse: Unified perception and prediction in birds-eye-view for vision-centric autonomous driving. arXiv preprint arXiv:2205.09743, 2022

    2022 arXiv

  51. [59]

    Fastrsr: Efficient and accurate road surface reconstruction from bird's eye view

    Zhao, Y., Ji, Y., Hao, X., and Li, S. Fastrsr: Efficient and accurate road surface reconstruction from bird's eye view. arXiv preprint arXiv:2504.09535, 2025

    2025 arXiv

  52. [60]

    a henb \

    Zhou, B. and Kr \"a henb \"u hl, P. Cross-view transformers for real-time map-view semantic segmentation. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 13760--13769, 2022

    2022

  53. [61]

    Himap: Hybrid representation learning for end-to-end vectorized hd map construction

    Zhou, Y., Zhang, H., Yu, J., Yang, Y., Jung, S., Park, S.-I., and Yoo, B. Himap: Hybrid representation learning for end-to-end vectorized hd map construction. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 15396--15406, 2024

    2024

  54. [62]

    Deformable DETR: deformable transformers for end-to-end object detection

    Zhu, X., Su, W., Lu, L., Li, B., Wang, X., and Dai, J. Deformable DETR: deformable transformers for end-to-end object detection. In ICLR . OpenReview.net, 2021

    2021

  55. [63]

    Understanding the robustness of 3d object detection with bird's-eye-view representations in autonomous driving

    Zhu, Z., Zhang, Y., Chen, H., Dong, Y., Zhao, S., Ding, W., Zhong, J., and Zheng, S. Understanding the robustness of 3d object detection with bird's-eye-view representations in autonomous driving. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 21600--2...

    2023

  56. [64]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

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