Pith. sign in

REVIEW 2 major objections 4 minor 64 references

Driver2Map: Imitating Human Driving for Online High-Definition Map Construction

T0 review · 2 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Driver2Map fuses onboard multi-view cameras, an SD map, and a satellite image to construct online HD maps, reporting state-of-the-art nuScenes results with mAP 62.9 and mIoU 54.0 under a 60 m × 30 m BEV range.

desk verdict The three-modality combination is genuinely new and the numbers look strong, but the paper never specifies how camera features get into BEV space—Eq. 4 sums image-plane features at identical raster positions, so the central module is geometrically ungrounded as written. read the letter →

arxiv 2608.01338 v1 pith:CUJY5MXL submitted 2026-08-02 cs.CV

Pan Yin , Runtian Xia , Weisong Kuang , Kaiyu Li , Cong Zhao , Xiangyong Cao This is my paper · ORCID
classification cs.CV
keywords onlineHDmapconstructionmulti-viewcamerafusionbird's-eye-viewperceptionSDpriorsatelliteimagemaskedautoencodernuScenesbenchmarkautonomousdriving
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

This paper argues that online high-definition map construction for autonomous driving improves when a model uses three complementary data sources at once—onboard multi-view cameras, a standard-definition map, and a satellite image—rather than the two used by prior systems. It introduces Driver2Map, which stages the fusion in the order a human driver builds a mental map: perceive the surroundings, focus on key road structure, add fine geometric detail, then use learned map priors to complete occluded regions. On the nuScenes benchmark the system reports the best published numbers, with mAP 62.9 and mIoU 54.0 at a 60 m × 30 m bird's-eye-view range. If the results hold, the remaining gap in online map prediction is less about sensing resolution and more about cross-modal alignment and occlusion reasoning.

What carries the argument

The central mechanism is Pose-Guided BEV Fusion (PGBF), which constructs a bird's-eye-view feature map as a normalized weighted sum of the six camera feature maps, with per-location weights derived from each camera's orientation, field of view, and distance to the BEV center. This carries the argument that cross-view interference can be suppressed without depth estimation or learned attention by letting each BEV location be dominated by its most relevant camera while keeping smooth transitions near view boundaries. Two supporting mechanisms complete the pipeline: a two-stage alignment procedure (an affine fit during data preparation followed by a learnable dense flow field inside the network) and Pretrained Prior for Map Refinement (PPMR), a masked autoencoder pretrained with category-wise masking and dynamic class weights to reconstruct occluded map structures.

What would settle it

A concrete check is to train the same pipeline but replace Equation 4 with a standard depth-based BEV lifting that projects each camera feature into the bird's-eye view before fusion. If that version matches or beats Driver2Map's 62.9 mAP and 54.0 mIoU, the claim that pose-guided weighting is what suppresses cross-view interference would be falsified.

Watch

Extended reading notes

Core claim

Driver2Map is an online HD map construction model that jointly exploits three modalities: six onboard camera images, an SD map, and a satellite image. The paper claims this triple fusion outperforms existing methods on nuScenes in both AP and IoU metrics, with the full system reaching mAP 62.9 and mIoU 54.0 under the 60 m × 30 m BEV range. The improvement is attributed to three components: a two-stage alignment strategy that first fits affine transforms between the nuScenes road map and each prior map and then learns a dense flow field to correct residual misalignment; a Pose-Guided BEV Fusion module that weights each camera's features by orientation and distance so each BEV location is dominated by its most relevant view; and a Pretrained Prior for Map Refinement module, a masked autoencoder that learns HD map structure and completes predictions under dynamic occlusion. The ablation study shows that removing Pose-Guided BEV Fusion causes the largest drop, from 54.0 to 44.0 mIoU, with the Crossing category affected most strongly, consistent with the claim that cross-view interference hurts elements visible from only one viewpoint.

Load-bearing premise

The load-bearing premise is that the six image-plane camera features can be treated as already living on the bird's-eye-view raster, so Equation 4 only needs to re-weight them at matching grid positions; no projection, depth estimation, or spatial cross-attention step is described between the image plane and the BEV grid.

Editorial extensions

If this is right

  • If the reported numbers hold, fusing SD and satellite priors together outperforms either prior alone, showing the two sources carry complementary rather than redundant information.
  • Removing Pose-Guided BEV Fusion costs 10.0 mIoU, the largest single ablation drop, so camera-pose weighting is a first-order component of the system's accuracy.
  • At 11.10 FPS on a single RTX 4090, the full three-modality model stays within the throughput range of published two-modality baselines, so the added priors do not force an offline pipeline.
  • The Crossing category shows the largest relative gain, consistent with the claim that cross-view suppression matters most for objects that appear in only one viewing direction.

Reading between the lines

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

  • A testable extension is to replace Equation 4's weighted sum with an explicit depth-based BEV lifting while keeping all other modules fixed; if the margin over the projection-free version vanishes, the reported gains come from pose weighting rather than from implicit geometry.
  • Because the PGBF weights are computed purely from camera geometry, the same module should transfer to any camera rig by updating calibration parameters, something the paper asserts but does not demonstrate on a second dataset.
  • The largest practical barrier to scaling is the per-city manual keypoint selection used for coarse alignment; an automated geo-referencing pipeline would be the natural next test of whether the method works beyond the regions used in the paper.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 4 minor

Summary. Driver2Map proposes an online HD map construction model for nuScenes that fuses three modalities: six onboard camera images, an SD map tile, and a satellite image tile. The method consists of three novel components: a two-stage alignment strategy for SD/satellite priors, a Pose-Guided BEV Fusion (PGBF) module that re-weights multi-view features using distance and orientation weights, and a Pretrained Prior for Map Refinement (PPMR) module based on masked autoencoding of HD maps. The paper reports state-of-the-art numbers, e.g., 54.0 mIoU and 62.9 mAP on the nuScenes validation set under a 60m x 30m BEV range, and attributes large ablation gains to PGBF.

Significance. The paper targets an important practical task, online HD map construction, and attempts to combine camera, SD-map, and satellite-image priors in one framework. If the technical proposal were sound, the reported gains on nuScenes would be a useful contribution, particularly the explicit handling of multi-camera overlap and dynamic occlusion. The manuscript also contains positive elements: it includes an efficiency comparison, extensive ablations, qualitative failure-case analysis, a self-critical limitations section, and a public code link. However, the central BEV-generation mechanism is not geometrically defined, and the reported numbers contain unexplained internal inconsistencies. As written, the key claim of a novel and effective image-to-BEV transformation cannot be verified or reproduced.

major comments (2)
  1. [3.2, Eq. (4), Figure 3] The PGBF module as written does not transform image features into BEV coordinates. The feature maps F_i^cam are described as the i-th camera's EfficientNet-B0 features with shape H x W x C, i.e., they live in the image plane, while the output F_bev is indexed by a BEV raster (x,y). Equation (4) is a weighted sum of the six feature maps at identical grid positions (x,y), modulated by distance and orientation weights. No projection, depth estimation, homography, or spatial cross-attention is defined between image coordinates and BEV coordinates. Since the six nuScenes cameras have different extrinsics, the same raster position in different camera feature maps corresponds to different physical ground locations, so Eq. (4) is an image-plane feature mixture, not a bird's-eye-view representation. This issue is load-bearing: Section 3.2 claims the module 'transform[s] onboard multi-view images into a BEV feature representation,' and Table 3 attributes a 10.0 mIoU and 5.4 mAP drop to PGBF. As written, the central mechanism is geometrically ungrounded and the headline results cannot be reproduced or assessed.
  2. [Tables 1 and 3] The AP numbers reported for the same configurations are internally inconsistent. Table 1 lists Driver2Map with C+SD+Sat at mAP 57.1, while Table 3, described as the same nuScenes validation set and the same 60m x 30m range, reports the full Driver2Map at mAP 62.9. Similarly, the C+Sat setting is 53.8 mAP in Table 1 but 60.2 mAP in Table 3 (the 'w/o SD' row). The IoU numbers also differ between Table 2 (C+Sat: 51.5 mIoU) and Table 3 (w/o SD: 50.0 mIoU). No explanation is given for these discrepancies, which undermines the credibility of the state-of-the-art comparison.
minor comments (4)
  1. [3.3] The text refers to 'Equation 3.3' when fitting the affine parameters a-f, but the affine equation is not numbered in the main text; please renumber or cite the appendix equation consistently.
  2. [4.3, Table 4] The alpha:beta ratio is selected as the best value on the validation set; this should be disclosed as a form of tuning on the evaluation target when interpreting the reported comparisons.
  3. [3.1] The notation H x W is used for both image feature maps and BEV rasters; please clarify the spatial resolution and coordinate conventions of each, since the relationship between them is central to the method.
  4. [Appendix D] The limitations discussion is honest and useful; however, the claim that PGBF can be 'directly adapted to different camera configurations' is not supported by the description, because no image-to-BEV geometric mapping is specified.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: supervised components and fitted hyperparameters do not make the central HD-map prediction equivalent to its inputs.

full rationale

Driver2Map's central claims are empirical: three supervised/trainable modules (PGBF, topology/detail enhancement, PPMR) are trained and evaluated on nuScenes, and the reported SOTA numbers are benchmark outcomes rather than restatements of any fitted quantity. The PPMR module pretrains an MAE on randomly masked nuScenes ground-truth maps and then fine-tunes to refine P_init; this is conventional supervised pretraining on target-domain labels, not a circular derivation. The two-stage alignment fits affine maps from manually selected keypoints and then crops prior tiles; the fitted parameters are data-preparation constants, and no final prediction is defined as those parameters. The alpha:beta ratio in Eq. 3 is selected on the validation set, a mild model-selection step, but the paper does not rename that choice as a prediction. The only self-citation, SamRoad++ (Yin et al. 2025), appears in related work and does not support any architectural decision, so it is not load-bearing. The serious concern in Eq. 4 — that F_bev is defined as a weighted sum of F_i^cam at identical raster indices without a specified image-to-BEV projection — is a geometric-grounding and correctness defect, not circularity, because the output is a weighted combination of inputs rather than an equivalence between a claimed prediction and a fitted input or a self-cited theorem.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim rests on three sets of assumptions: the BEV fusion coordinate equivalence (Eq. 4), the planar and sector camera model in PGBF, and the transferability of MAE-learned map priors. The affine alignment parameters and alpha:beta are fitted or tuned rather than derived, but they are standard calibration-style parameters. No new physical entities are introduced.

free parameters (3)
  • alpha and beta in Eq. 3 = alpha:beta = 1:2
    Balancing coefficients for distance vs orientation weights; selected by validation-set sweep (Table 4), not derived.
  • Affine parameters a-f for SD and satellite alignment = per-city least-squares fits (e.g., x_sd' = 0.4517x + 0.0001y - 611.09)
    Estimated from N=50 manually selected keypoint pairs per nuScenes region (Algorithm 1); these are fitted to prior map correspondences, not to target labels.
  • N keypoint pairs = 50
    Manual keypoint count chosen for each city; no sensitivity analysis.
assumptions (4)
  • ad hoc to paper Camera image feature maps F_i^cam and BEV raster (x,y) share the same coordinate grid (Eq. 4).
    No projection from image plane to BEV is provided; PGBF weights are defined on BEV coordinates but multiplied with features indexed at identical grid positions.
  • domain assumption Ground plane is approximately flat and all cameras share the ego center, so angular sectors around the ego define useful camera-to-BEV correspondence.
    Section 3.2 partitions BEV into sectors by camera orientation and FOV; camera height, perspective, and radial distortion are ignored.
  • domain assumption Residual misalignment after affine coarse alignment is small and can be corrected by the learned flow field in fine alignment.
    Section 3.3 and Eq. 5 assume pixelwise flow warping can absorb residual map offsets.
  • domain assumption MAE pretraining on nuScenes ground-truth HD maps transfers to occluded regions at inference.
    Section 3.4; an empirical claim not demonstrated beyond the same-dataset validation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Driver2Map: Imitating Human Driving for Online High-Definition Map Construction." pith.science (2026). https://pith.science/paper/CUJY5MXL

@misc{pith2026260801338,
  author       = {Pith},
  title        = {Pith review of: Driver2Map: Imitating Human Driving for Online High-Definition Map Construction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CUJY5MXL}},
  note         = {Machine review of arXiv:2608.01338}
}
read the original abstract

High-definition (HD) maps are essential for autonomous driving systems. In constructing such maps, onboard multi-view camera images, standard-definition maps and satellite images provide crucial information. However, due to the modality and perspective differences among these data sources, existing methods often struggle to effectively align and fuse them, making online HD map construction still challenging. To address these issues, we propose Driver2Map, an online HD map construction model inspired by human drivers. Unlike existing HD map construction models that utilize only two modalities, our Driver2Map can simultaneously exploit three modalities. Specifically, we propose a "two-stage alignment" strategy to reduce spatial misalignment across different modalities. Additionally, we introduce "Pose-Guided BEV Fusion", a BEV (bird's-eye-view) generation module that leverages camera pose information to adaptively weight multi-view features, thereby effectively suppressing cross-view feature overlap during BEV generation. Also, we design a "Pretrained Prior for Map Refinement" module to refine the initial prediction by learning map structure priors, thus improving the HD map prediction under dynamic occlusions. Extensive experiments demonstrate that Driver2Map outperforms existing methods on both IoU and AP metrics.

Figures

Figures reproduced from arXiv: 2608.01338 by the authors.

Figure 1
Figure 1. Comparison between our Driver2Map and other [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Motivation of Driver2Map: multi data source limitations (left) and human drivers’ map-building process (right). The [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Overview of Driver2Map. Inspired by the process of map construction by human drivers, our model follows four [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Visualization of offsets across different maps and [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: We compare our predictions with other methods within a [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Visualization of the six camera-weight maps generated by PGBF for a representative sample from the nuScenes dataset. [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: Detailed workflow for generating coarsely aligned SD map and satellite image priors, illustrated using Boston Seaport. [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: Additional qualitative results on the nuScenes validation set. Columns show onboard surround-view camera images, [PITH_FULL_IMAGE:figures/full_fig_p013_8.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

64 extracted references · 49 canonical work pages

  1. [1]

    IEEE Transactions on Robotics , volume=

    ORB-SLAM2: An Open-Source SLAM System for Monocular, Stereo, and RGB-D Cameras , author=. IEEE Transactions on Robotics , volume=

    work page

  2. [2]

    2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages=

    LIO-SAM: Tightly-coupled Lidar Inertial Odometry via Smoothing and Mapping , author=. 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages=. 2020 , organization=

    work page 2020

  3. [3]

    2022 International Conference on Robotics and Automation (ICRA) , pages=

    Hdmapnet: An online hd map construction and evaluation framework , author=. 2022 International Conference on Robotics and Automation (ICRA) , pages=. 2022 , organization=

    work page 2022

  4. [4]

    International Conference on Machine Learning (ICML) , pages=

    Vectormapnet: End-to-end vectorized hd map learning , author=. International Conference on Machine Learning (ICML) , pages=. 2023 , organization=

    work page 2023

  5. [5]

    International Conference on Learning Representations , year=

    MapTR: Structured Modeling and Learning for Online Vectorized HD Map Construction , author=. International Conference on Learning Representations , year=

    work page

  6. [6]

    IEEE Robotics and Automation Letters (IRAL) , volume=

    P-mapnet: Far-seeing map generator enhanced by both sdmap and hdmap priors , author=. IEEE Robotics and Automation Letters (IRAL) , volume=. 2024 , publisher=

    work page 2024

  7. [7]

    European Conference on Computer Vision , pages=

    Enhancing vectorized map perception with historical rasterized maps , author=. European Conference on Computer Vision , pages=. 2024 , organization=

    work page 2024

  8. [8]

    arXiv preprint arXiv:2411.14751 , year=

    Toposd: Topology-enhanced lane segment perception with sdmap prior , author=. arXiv preprint arXiv:2411.14751 , year=

    arXiv

Show all 64 references
  1. [9]

    2024 IEEE International Conference on Robotics and Automation (ICRA) , pages=

    Complementing onboard sensors with satellite maps: a new perspective for hd map construction , author=. 2024 IEEE International Conference on Robotics and Automation (ICRA) , pages=. 2024 , organization=

    2024

  2. [10]

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

    NuScenes: A multimodal dataset for autonomous driving , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=

  3. [11]

    IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=

    Argoverse: 3D Tracking and Forecasting with Rich Maps , author=. IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=

  4. [12]

    2016 13th International Conference on Ubiquitous Robots and Ambient Intelligence (URAI) , pages=

    Adaptive inverse perspective mapping for lane map generation with SLAM , author=. 2016 13th International Conference on Ubiquitous Robots and Ambient Intelligence (URAI) , pages=. 2016 , organization=

    2016

  5. [13]

    2023 IEEE international conference on robotics and automation (ICRA) , pages=

    Bevfusion: Multi-task multi-sensor fusion with unified bird's-eye view representation , author=. 2023 IEEE international conference on robotics and automation (ICRA) , pages=. 2023 , organization=

    2023

  6. [14]

    Advances in Neural Information Processing Systems , volume=

    BEVFusion: A Simple and Robust LiDAR-Camera Fusion Framework , author=. Advances in Neural Information Processing Systems , volume=

  7. [15]

    European Conference on Computer Vision (ECCV) , year=

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

  8. [16]

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

    Streammapnet: Streaming mapping network for vectorized online hd map construction , author=. Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , pages=

  9. [17]

    Advances in Neural Information Processing Systems , volume=

    Online map vectorization for autonomous driving: A rasterization perspective , author=. Advances in Neural Information Processing Systems , volume=

  10. [18]

    European conference on computer vision (ECCV) , pages=

    Lift, splat, shoot: Encoding images from arbitrary camera rigs by implicitly unprojecting to 3d , author=. European conference on computer vision (ECCV) , pages=. 2020 , organization=

    2020

  11. [19]

    2025 IEEE Intelligent Vehicles Symposium (IV) , pages=

    SAM-Maps: Road Map Generation for Automated Vehicles in Urban Areas , author=. 2025 IEEE Intelligent Vehicles Symposium (IV) , pages=. 2025 , organization=

    2025

  12. [20]

    Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) , pages=

    Towards satellite image road graph extraction: A global-scale dataset and a novel method , author=. Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) , pages=

  13. [21]

    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=

  14. [22]

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

    Segment anything model for road network graph extraction , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  15. [23]

    International journal of computer applications , volume=

    Satellite image classification methods and techniques: A review , author=. International journal of computer applications , volume=. 2015 , publisher=

    2015

  16. [24]

    Science , volume=

    Using satellite imagery to understand and promote sustainable development , author=. Science , volume=. 2021 , publisher=

    2021

  17. [25]

    Information Fusion , volume=

    An integrated system for automatic road mapping from high-resolution multi-spectral satellite imagery by information fusion , author=. Information Fusion , volume=. 2005 , publisher=

    2005

  18. [26]

    GeoInformatica , volume=

    Automatically conflating road vector data with orthoimagery , author=. GeoInformatica , volume=. 2006 , publisher=

    2006

  19. [27]

    International Conference on Machine Learning , pages=

    Learning Transferable Visual Models From Natural Language Supervision , author=. International Conference on Machine Learning , pages=

  20. [28]

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

    Masked Autoencoders Are Scalable Vision Learners , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=

  21. [29]

    5 at micro scale , author=

    Spatial-temporal heterogeneity of air pollution: The relationship between built environment and on-road PM2. 5 at micro scale , author=. Transportation Research Part D: Transport and Environment , volume=. 2019 , publisher=

    2019

  22. [30]

    Machines , volume=

    Localization and mapping for self-driving vehicles: A survey , author=. Machines , volume=. 2024 , publisher=

    2024

  23. [31]

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

    Simmim: A simple framework for masked image modeling , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  24. [32]

    arXiv preprint arXiv:2112.11790 , year=

    BEVDet: High-performance Multi-camera 3D Object Detection in Bird-Eye-View , author=. arXiv preprint arXiv:2112.11790 , year=

    arXiv

  25. [33]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Bevdepth: Acquisition of reliable depth for multi-view 3d object detection , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  26. [34]

    Proceedings of the AAAI conference on artificial intelligence , volume=

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

  27. [35]

    Proceedings of the IEEE/CVF international conference on computer vision , pages=

    Sparsebev: High-performance sparse 3d object detection from multi-camera videos , author=. Proceedings of the IEEE/CVF international conference on computer vision , pages=

  28. [36]

    arXiv preprint arXiv:2301.07870 , year=

    Fast-BEV: Towards real-time on-vehicle bird's-eye view perception , author=. arXiv preprint arXiv:2301.07870 , year=

    arXiv

  29. [37]

    International conference on machine learning , pages=

    Efficientnet: Rethinking model scaling for convolutional neural networks , author=. International conference on machine learning , pages=. 2019 , organization=

    2019

  30. [38]

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

    Deep residual learning for image recognition , author=. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , pages=

  31. [39]

    International Conference on Medical image computing and computer-assisted intervention , pages=

    U-net: Convolutional networks for biomedical image segmentation , author=. International Conference on Medical image computing and computer-assisted intervention , pages=. 2015 , organization=

    2015

  32. [40]

    International Conference on Learning Representations (ICLR) , year=

    An image is worth 16x16 words: Transformers for image recognition at scale , author=. International Conference on Learning Representations (ICLR) , year=

  33. [41]

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

    Generalized Intersection over Union: A Metric and A Loss for Bounding Box Regression , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=

  34. [42]

    2020 IEEE international conference on robotics and automation (ICRA) , pages=

    UrbanLoco: A full sensor suite dataset for mapping and localization in urban scenes , author=. 2020 IEEE international conference on robotics and automation (ICRA) , pages=. 2020 , organization=

    2020

  35. [43]

    International Journal of Computer Vision (IJCV) , volume=

    Maptrv2: An end-to-end framework for online vectorized hd map construction , author=. International Journal of Computer Vision (IJCV) , volume=. 2025 , publisher=

    2025

  36. [44]

    2024 International Conference on 3D Vision (3DV) , pages=

    PIVOT-Net: Heterogeneous point-voxel-tree-based framework for point cloud compression , author=. 2024 International Conference on 3D Vision (3DV) , pages=. 2024 , organization=

    2024

  37. [45]

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

    Pivotnet: Vectorized pivot learning for end-to-end hd map construction , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  38. [46]

    arXiv preprint arXiv:2503.05492 , year=

    Fastmap: fast queries initialization based vectorized hd map reconstruction framework , author=. arXiv preprint arXiv:2503.05492 , year=

    arXiv

  39. [47]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Bev-mae: Bird’s eye view masked autoencoders for point cloud pre-training in autonomous driving scenarios , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  40. [48]

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

    Sg-bev: Satellite-guided bev fusion for cross-view semantic segmentation , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=

  41. [49]

    Applied Sciences , volume=

    Loop-MapNet: A Multi-Modal HDMap Perception Framework with SDMap Dynamic Evolution and Priors , author=. Applied Sciences , volume=. 2025 , publisher=

    2025

  42. [50]

    arXiv preprint arXiv:2601.10512 , year=

    SatMap: Revisiting Satellite Maps as Prior for Online HD Map Construction , author=. arXiv preprint arXiv:2601.10512 , year=

  43. [51]

    arXiv preprint arXiv:2512.11319 , year=

    SATMapTR: Satellite Image Enhanced Online HD Map Construction , author=. arXiv preprint arXiv:2512.11319 , year=

  44. [52]

    arXiv preprint arXiv:1909.05003 , year=

    Human visual attention prediction boosts learning & performance of autonomous driving agents , author=. arXiv preprint arXiv:1909.05003 , year=

    1909 arXiv

  45. [53]

    Robotics and Autonomous Systems , volume=

    Static map generation from 3D LiDAR point clouds exploiting ground segmentation , author=. Robotics and Autonomous Systems , volume=. 2023 , publisher=

    2023

  46. [54]

    Proceedings of the IEEE international conference on computer vision (ICCV) , pages=

    Deformable convolutional networks , author=. Proceedings of the IEEE international conference on computer vision (ICCV) , pages=

  47. [55]

    European conference on computer vision , pages=

    Deformable feature aggregation for dynamic multi-modal 3D object detection , author=. European conference on computer vision , pages=. 2022 , organization=

    2022

  48. [56]

    2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages=

    Hierarchical road topology learning for urban mapless driving , author=. 2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages=. 2022 , organization=

    2022

  49. [57]

    IEEE transactions on visualization and computer graphics , volume=

    Drawing and labeling high-quality metro maps by mixed-integer programming , author=. IEEE transactions on visualization and computer graphics , volume=. 2010 , publisher=

    2010

  50. [58]

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

    Flownet 2.0: Evolution of optical flow estimation with deep networks , author=. Proceedings of the IEEE conference on computer vision and pattern recognition (CVPR) , pages=

  51. [59]

    Advances in neural information processing systems (Neurips) , volume=

    Spatial transformer networks , author=. Advances in neural information processing systems (Neurips) , volume=

  52. [60]

    Advances in Neural Information Processing Systems , volume=

    Sdtagnet: Leveraging text-annotated navigation maps for online hd map construction , author=. Advances in Neural Information Processing Systems , volume=

  53. [61]

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

    DAMap: Distance-aware MapNet for High Quality HD Map Construction , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , pages=

  54. [62]

    2024 IEEE International Conference on Robotics and Automation (ICRA) , pages=

    Superfusion: Multilevel lidar-camera fusion for long-range hd map generation , author=. 2024 IEEE International Conference on Robotics and Automation (ICRA) , pages=. 2024 , organization=

    2024

  55. [63]

    Scientific Reports , year=

    CRFusion: a novel LiDAR-camera fusion network for BEV map construction , author=. Scientific Reports , year=

  56. [64]

    Remote Sensing , volume=

    Geolocalization from aerial sensing images using road network alignment , author=. Remote Sensing , volume=. 2024 , publisher=

    2024

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.