Pith. sign in

REVIEW 4 major objections 5 minor 2 cited by

MapFusion: A Novel BEV Feature Fusion Network for Multi-modal Map Construction

T0 review · 4 major / 5 minor · reviewed 2026-08-09 · deepseek-v4-flash

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

Pith's one-line read The paper proposes MapFusion, a plug-and-play camera-LiDAR BEV fusion scheme whose Cross-modal Interaction Transform and Dual Dynamic Fusion modules improve HD map construction mAP by 3.6% and BEV map segmentation mIoU by 6.2% on nuScenes.

desk verdict Useful fusion recipe with consistent ablations, but the HD-map headline gain rests on a MapTR baseline the cited paper never published. read the letter →

arxiv 2502.04377 v1 pith:3EG7IMWQ submitted 2025-02-05 cs.CV cs.AI

Xiaoshuai Hao , Yunfeng Diao , Mengchuan Wei , Yifan Yang , Peng Hao , Rong Yin , Hui Zhang , Weiming Li
show 2 more authors
Shu Zhao Yu Liu
This is my paper · ORCID
classification cs.CVcs.AI
keywords BEVfeaturefusionmulti-modalmapconstructioncamera-LiDARcross-modalinteractionself-attentionHDsegmentationautonomousdriving
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

MapFusion is a camera-LiDAR feature fusion method for the two main autonomous-driving map tasks: vectorized high-definition map construction and bird's-eye-view (BEV) map segmentation. The paper's argument is that projecting camera and LiDAR features into the same BEV grid does not remove their semantic misalignment, and that the element-wise addition, weighted averaging, or concatenation used by previous fusion methods leaves that misalignment uncorrected. MapFusion therefore inserts two plug-in modules between the modality encoders and the map decoder: a Cross-modal Interaction Transform (CIT) that runs multi-head self-attention over the concatenated camera and LiDAR BEV tokens, and a Dual Dynamic Fusion (DDF) module that gates the two feature maps with complementary attention weights before merging them. Reported results on nuScenes show absolute gains of 3.6% mAP for HD map construction and 6.2% mIoU for BEV map segmentation over the corresponding baselines, with similar gains on Argoverse2, and the modules also improve HDMapNet, VectorMapNet, and MapTR when added on top. The paper concludes that explicitly aligning the two modalities inside the fusion stage is worth more than the naive fusion operations that current pipelines use.

What carries the argument

The load-bearing machinery is the CIT-DDF pair inserted between the modality encoders and the map decoder. CIT flattens the camera and LiDAR BEV features into tokens $\mathbf{T}^{BEV}_{Camera}$ and $\mathbf{T}^{BEV}_{LiDAR}$, concatenates them with a learnable positional embedding, and applies multi-head self-attention; the correlation matrix $\alpha$ splits into intra-modality and inter-modality blocks, so the self-attention simultaneously aligns each modality with itself and with the other. DDF takes the CIT outputs, generates a channel attention vector $\mathbf{w}$ from their sum via global average pooling and linear layers with a sigmoid, multiplies the camera features by $\mathbf{w}$ and the LiDAR features by $1-\mathbf{w}$, concatenates them, and applies a $3\times3$ convolution followed by an adaptive channel-spatial gate. This mechanism carries the paper's claim: CIT removes the semantic misalignment that naive fusion leaves in place, and DDF prevents information loss by softly selecting which modality to trust at each channel.

What would settle it

Retrain MapTR and BEVFusion under MapFusion's exact settings, including the same epochs, batch size, learning rate, and data pre-processing, and compare mAP and mIoU; if the gap drops below roughly one point or disappears, the improvements are not caused by the CIT and DDF modules.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that the residual error in multi-modal map construction after BEV fusion is not a resolution or capacity problem but a correspondence problem: camera and LiDAR features at the same BEV grid location can represent different semantic content because the view transformation is depth-imperfect and the sensor modalities differ. The paper's solution, the Cross-modal Interaction Transform (CIT), treats the two feature maps as one token sequence and learns a correlation matrix $\alpha = \mathrm{softmax}(\mathbf{Q}\mathbf{K}^{\top}/\sqrt{D_k})$ whose four natural blocks are camera-to-camera, LiDAR-to-LiDAR, camera-to-LiDAR, and LiDAR-to-camera attention. This lets each spatial position be re-weighted by evidence from the other modality before fusion, which the paper demonstrates with t-SNE visualizations showing the two feature populations moving closer together after CIT. The Dual Dynamic Fusion (DDF) module then computes a channel-wise gate from the summed features and applies it to the two inputs in complementary form ($\mathbf{w}$ and $1-\mathbf{w}$), so the network can suppress the weaker modality where the stronger one is informative. Together these modules form MapFusion, and the paper reports that this interaction-then-selection design outperforms its baselines by the margins above on two tasks and two datasets.

Load-bearing premise

The reported gains assume that MapTR, BEVFusion, and the re-implemented Argoverse2 baselines were trained with the same backbone, schedule, augmentation, and evaluation protocol as MapFusion, so the 3.6 to 6.5 point differences come from the proposed modules rather than from training disparities.

Editorial extensions

If this is right

  • On nuScenes HD map construction, MapFusion reaches 66.1 mAP, beating MapTR's 62.5, with the largest single-class gain on pedestrian crossings (61.6 vs 55.9).
  • On nuScenes BEV map segmentation, MapFusion reaches 68.9 mIoU, beating BEVFusion's 62.7, and improves every class except car park.
  • The improvements transfer to Argoverse2: +4.1 mAP for HD map construction and +6.5 mIoU for BEV map segmentation over re-implemented baselines.
  • Adding MapFusion to re-implemented HDMapNet and VectorMapNet and to MapTR yields +7.6, +5.5, and +3.6 mAP respectively on nuScenes.
  • Ablations attribute the larger part of the segmentation gain to CIT (+5.6 mIoU alone) and a smaller part to DDF (+1.4), with the full model combining both.

Reading between the lines

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

  • Because CIT is defined purely on BEV tokens, the same alignment mechanism could transfer to other multi-modal BEV tasks such as 3D object detection or occupancy prediction; the paper states such an extension as future work without testing it.
  • A per-condition evaluation could test whether CIT's benefit grows as camera depth estimates become less reliable, since the paper's misalignment story predicts the largest gains where the camera-to-BEV projection is weakest.
  • DDF's complementary gate could be refined from channel-level weights to per-location routing, letting the network decide which modality owns each BEV cell; that is a natural next step the paper leaves implicit.
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

4 major / 5 minor

Summary. The paper proposes MapFusion, a camera-LiDAR BEV feature fusion method for map construction, consisting of two modules: a Cross-modal Interaction Transform (CIT) that applies multi-head self-attention to the concatenated camera and LiDAR BEV tokens, and a Dual Dynamic Fusion (DDF) module that uses gated weighting and squeeze-excitation to fuse the two feature streams. The authors evaluate MapFusion on HD map construction and BEV map segmentation on nuScenes and Argoverse2, reporting absolute improvements of +3.6 mAP, +4.1 mAP, +6.2 mIoU, and +6.5 mIoU over their baselines, along with ablation studies, compatibility experiments, and visualizations. The paper is an extension of the authors' ICRA 2024 conference paper and adds the BEV segmentation task, additional ablations, and further analysis.

Significance. If the reported results are reproducible, the proposed modules constitute a simple and potentially useful plug-and-play contribution to multi-modal BEV fusion for map-related tasks. The compatibility study in Table 9 and the t-SNE and feature-map visualizations provide supporting evidence that the modules can be inserted into existing pipelines. However, the evaluation infrastructure currently prevents acceptance of the headline claims: the provenance of the key HD-map baseline is questionable, there is an internal backbone inconsistency in the BEV segmentation ablation, and all results are single runs without variance information. The method itself is not mathematically derived, but the empirical claims are the main contribution, so these evaluation issues are load-bearing.

major comments (4)
  1. [Table 1 and Section 2.1] The caption of Table 1 states that the comparison numbers are 'taken from MapTR [25]', but MapTR (ICLR 2023) is a camera-only method; the LiDAR-only (55.6 mAP) and camera-LiDAR (62.5 mAP) rows cannot come from that publication. Section 2.1 also lists MapTR among 'camera-LiDAR fusion methods', which is a misclassification. Because the headline +3.6 mAP improvement on HD map construction is measured against this unvalidated control, the central claim is not adequately supported. Please provide the exact origin of the LiDAR and C&L MapTR numbers, specify how those variants were implemented (including the fusion operation and training schedule), and release code or model weights so the control can be reproduced.
  2. [Table 6 vs. Table 3 and Section 4.3] The BEV segmentation ablation in Table 6 labels the backbone as 'ResNet-50 & VoxelNet', while Table 3 and Section 4.3 state that the BEV segmentation experiments use Swin-T and VoxelNet. Moreover, the baseline row in Table 6 reports the same mIoU (62.7) as the Swin-T baseline in Table 3, which cannot be true if the backbone differs. This internal inconsistency affects the interpretation of the +6.2 mIoU improvement and the component ablation. Please clarify the actual backbone and report a single consistent set of results.
  3. [Tables 5-9] All results are reported as single runs with no variance information. Several conclusions hinge on small differences, e.g., Add Fusion vs. Dynamic Fusion on HD map construction (64.4 vs. 64.3 mAP in Table 7) and Add Fusion vs. Baseline on BEV segmentation (+0.2 mIoU in Table 8). Without multi-seed statistics, these differences and even some of the claimed component gains may not be statistically meaningful. Please report mean and standard deviation over at least three seeds for the main results and ablations.
  4. [Tables 2, 4, and 9] The Argoverse2 results (Tables 2 and 4) and the compatibility results (Table 9) rely on author re-implementations marked with †, with no code release or detailed configuration files. The HDMapNet† re-implementation in Table 9 (28.2 mAP) is below the published HDMapNet C&L result (31.0 mAP in Table 1), which suggests the re-implemented controls may be systematically under-tuned. Please provide the exact training configs, logs, and code for the re-implemented baselines, and compare with any available published numbers.
minor comments (5)
  1. [Section 4.3] The sentence 'Training losses include classification loss, point2point loss, and edge direction loss. with weights of 2.0, 5.0, and 0.005' has a punctuation error, and the learning-rate description 'starting for 1e-4 and performing a single cycle with target ratios 10, 1e-4 and a step of 0.4' is unclear. Please rewrite for precision.
  2. [Equation 5] The correlation matrix in Eq. 5 has misaligned row and column labels and is difficult to read. Please reformat the matrix so that the intra-modality and inter-modality blocks are clearly labeled.
  3. [Table 3] The M2BEV row contains '%' placeholders for several classes. Please either fill in the actual values or remove the row to avoid presenting incomplete data.
  4. [Section 3.4 remarks] The description of the AvgPool operations is confusing: Eq. 6 says AvgPool outputs a vector of length C, while the remarks say the AvgPool in Eq. 8 outputs a W x H map, and Eq. 8 then applies a linear transform W to that map. Please clarify the exact tensor shapes and the role of the linear transform.
  5. [Section 5 and Conclusion] The conclusion repeatedly describes CIT as a 'cross-attention mechanism', but Section 3.3 defines it as self-attention over the concatenated BEV tokens. Please align the terminology throughout the paper.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the reported gains are empirical measurements, not derivations that reduce to the method's own definitions or to a fitted parameter.

full rationale

The paper's central claims are experimental: MapFusion's CIT and DDF modules are inserted into baseline pipelines and the resulting mAP/mIoU values are measured on nuScenes and Argoverse2. No equation defines the reported metric in terms of a fitted parameter, and no 'prediction' is obtained by construction from the training objective. The ablations in Tables 5-8 compare trained model variants, so the contribution of each module is an empirical outcome rather than a tautology. The self-citation of the authors' ICRA 2024 MBFusion paper is an extension statement, not a load-bearing justification of correctness; the modules are validated by the paper's own controlled experiments and against external baselines. There is no imported uniqueness theorem and no ansatz smuggled in via citation. One caveat worth flagging for correctness risk rather than circularity: Table 1 attributes the MapTR LiDAR-only and camera+LiDAR rows to MapTR [25], but MapTR is a camera-only method in its original publication, and Section 2.1 likewise misclassifies MapTR as a camera-LiDAR fusion method. That is a benchmark-provenance and external-validity concern that could affect the strength of the +3.6 comparison, but it does not make the derivation circular: the baselines are external measured numbers, not quantities produced by the paper's own equations or by fitting the method to the target metric. On the circularity axis, the derivation chain is self-contained and no step reduces to its own input.

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

The central claim depends on learned weights in the attention and gating modules, which are trained on the standard datasets rather than fitted to the target metrics. No hand-set constants are used to produce the reported mAP/mIoU values. The evaluation, however, depends on the fairness of the baseline comparisons and on the standard dataset protocols.

assumptions (3)
  • domain assumption Self-attention over concatenated camera and LiDAR BEV tokens can reduce semantic misalignment and improve downstream map construction performance.
    This is the core modeling assumption behind the CIT module (Section 3.3). It is plausible given prior cross-modal attention results, but it is not proven independently of the reported experiments.
  • domain assumption The nuScenes and Argoverse2 datasets, with the standard splits and evaluation metrics (mAP with Chamfer thresholds, mIoU with binary per-class evaluation), provide a fair and representative benchmark for map construction methods.
    The paper follows the evaluation protocol of MapTR and BEVFusion (Section 4.1-4.2). The validity of the comparison depends on these datasets being appropriate proxies for real-world map construction.
  • ad hoc to paper The baseline numbers taken from MapTR [25] and BEVFusion [31], and the re-implemented baselines (marked †), were produced under settings equivalent to MapFusion's (same backbone, training schedule, data preprocessing).
    The paper states 'the numbers are taken from MapTR' and 'we re-implemented the BEVFusion method following the original settings' (Section 4.4). If the control conditions differ, the reported gains could be due to training setup rather than the modules.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MapFusion: A Novel BEV Feature Fusion Network for Multi-modal Map Construction." pith.science (2026). https://pith.science/paper/3EG7IMWQ

@misc{pith2026250204377,
  author       = {Pith},
  title        = {Pith review of: MapFusion: A Novel BEV Feature Fusion Network for Multi-modal Map Construction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3EG7IMWQ}},
  note         = {Machine review of arXiv:2502.04377}
}
read the original abstract

Map construction task plays a vital role in providing precise and comprehensive static environmental information essential for autonomous driving systems. Primary sensors include cameras and LiDAR, with configurations varying between camera-only, LiDAR-only, or camera-LiDAR fusion, based on cost-performance considerations. While fusion-based methods typically perform best, existing approaches often neglect modality interaction and rely on simple fusion strategies, which suffer from the problems of misalignment and information loss. To address these issues, we propose MapFusion, a novel multi-modal Bird's-Eye View (BEV) feature fusion method for map construction. Specifically, to solve the semantic misalignment problem between camera and LiDAR BEV features, we introduce the Cross-modal Interaction Transform (CIT) module, enabling interaction between two BEV feature spaces and enhancing feature representation through a self-attention mechanism. Additionally, we propose an effective Dual Dynamic Fusion (DDF) module to adaptively select valuable information from different modalities, which can take full advantage of the inherent information between different modalities. Moreover, MapFusion is designed to be simple and plug-and-play, easily integrated into existing pipelines. We evaluate MapFusion on two map construction tasks, including High-definition (HD) map and BEV map segmentation, to show its versatility and effectiveness. Compared with the state-of-the-art methods, MapFusion achieves 3.6% and 6.2% absolute improvements on the HD map construction and BEV map segmentation tasks on the nuScenes dataset, respectively, demonstrating the superiority of our approach.

Figures

Figures reproduced from arXiv: 2502.04377 by the authors.

Figure 1
Figure 1. Illustration of different modalities BEV features. Although both LiDAR and camera BEV features are presented in the shared BEV space, they may still be semantically misaligned due to the significant modality gap. (Best viewed in color. Blue color means small values and red means large.) semantic information, but methods relying solely on them often struggle with spatial distortions when projecting Perspective View (… view at source ↗
Figure 2
Figure 2. Illustration of different map construction tasks (HD map construction and BEV map segmentation). 2. Related Work Our work is highly related to map construction task (See [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. An overview of MapFusion framework. First, we extract features from multi-modal inputs and convert them into a shared bird’s-eye view (BEV) space efficiently using view transformations. To fuse the BEV features from different modalities, we first propose Cross-modal Interaction Transform (CIT) module to enhance one modality from another modality by self-attention mechanism. Afterwards, we propose a Dual Dynamic Fusi… view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Illustration of the Correlation Matrix 𝛼. to get the input BEV tokens 𝐓 in ∈ ℝ2𝐻𝑊 ×𝐶 of the Transformer [43]. The positional embedding enables the model to differentiate spatial information between different tokens at training time. Third, the input token 𝐓 in uses lin…
Figure 5
Figure 5. Figure 5: Three existing fusion strategies and our proposed Dual Dynamic Fusion (DDF) strategy. 3.5. Map-Task Heads We apply specific heads for different map tasks to the fused BEV features. We show two examples: HD map construction and BEV map segmentation. HD map construction …
Figure 6
Figure 6. Figure 6: Accuracy-Computation Analysis. We report the accuracy-computation trade-off by utilizing our proposed cross￾modal interaction transform module and different fusion strategies. (a) Before CIT Module (b) After CIT Module [PITH_FULL_IMAGE:figures/full_fig_p015_6.png]
Figure 7
Figure 7. Figure 7: The t-SNE visualizations of (a) Before CIT module and (b) After CIT module on HD map construction task. Red/Blue denotes camera BEV feature/LiDAR BEV feature. After the CIT module, the BEV features from different modalities are aligned in a shared space, i.e, red and b…
Figure 8
Figure 8. Figure 8: Visualization of feature maps before and after the CIT module for the HD map construction task. construction task, showing both LiDAR and camera inputs. We compare the predicted vectorized HD map results of different models, including HDMapNet [21], VectorMapNet [28], …
Figure 9
Figure 9. Figure 9: Qualitative results on BEV map segmentation task. We present a sample scene from nuScenes: a) six camera inputs, b) LiDAR scan, c) ground-truth BEV map segmentation map, d) baseline BEV segmentation map (BEVFusion [31]), e) BEV segmentation map of only using CIT module…
Figure 10
Figure 10. Figure 10: Qualitative results on HD map task. We present a sample scene from nuScenes: a) six camera inputs, b) LiDAR scan, c) ground-truth BEV vectorized HD map, d) HDMapNet [21], e) VectorMapNet [28], f) baseline BEV vectorized HD map (MapTR [25]), g) BEV vectorized HD map of…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

  1. Affine Modulation-based Audiogram Fusion Network for Joint Noise Reduction and Hearing Loss Compensation

    eess.AS 2025-09 conditional novelty 6.0 of 10

    A hearing-aid network that injects the user's audiogram into a speech-enhancement model with affine modulation beats existing joint noise-reduction and compensation systems on objective quality metrics.

  2. Progressive Bird's Eye View Perception for Safety-Critical Autonomous Driving: A Comprehensive Survey

    cs.RO 2025-08 conditional novelty 5.0 of 10

    A safety-critical survey that organizes BEV perception into single-modality, multimodal, and collaborative stages and consolidates robustness evidence that multimodal fusion degrades far less than single-modality perc...

Reference graph

Works this paper leans on

59 extracted references · 51 canonical work pages · cited by 2 Pith papers

  1. [25]

    Maptr:Structuredmodelingandlearningforonlinevectorized HD map construction, in: International Conference on Learning Representations

    Liao,B.,Chen,S.,Wang,X.,Cheng,T.,Zhang,Q.,Liu,W.,Huang,C.,2023a. Maptr:Structuredmodelingandlearningforonlinevectorized HD map construction, in: International Conference on Learning Representations

    work page

  2. [1]

    A geometric approach to obtain a bird’s eye view from an image, in: Proceedings of the IEEE/CVF international conference on computer vision workshops

    Ammar Abbas, S., Zisserman, A., 2019. A geometric approach to obtain a bird’s eye view from an image, in: Proceedings of the IEEE/CVF international conference on computer vision workshops

    work page 2019

  3. [2]

    X-align: Cross-modal cross-view alignment forbird’s-eye-viewsegmentation,in:ProceedingsoftheIEEE/CVFWinterConferenceonApplicationsofComputerVision,pp.3287–3297

    Borse, S., Klingner, M., Kumar, V.R., Cai, H., Almuzairee, A., Yogamani, S., Porikli, F., 2023. X-align: Cross-modal cross-view alignment forbird’s-eye-viewsegmentation,in:ProceedingsoftheIEEE/CVFWinterConferenceonApplicationsofComputerVision,pp.3287–3297

    work page 2023

  4. [3]

    11618–11628

    Caesar,H.,Bankiti,V.,Lang,A.H.,Vora,S.,Liong,V.E.,Xu,Q.,Krishnan,A.,Pan,Y.,Baldan,G.,Beijbom,O.,2020.nuscenes:Amultimodal dataset for autonomous driving, in: IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 11618–11628

    work page 2020

  5. [4]

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

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

    arXiv

  6. [5]

    Focal sparse convolutional networks for 3d object detection, in: IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Chen, Y., Li, Y., Zhang, X., Sun, J., Jia, J., 2022b. Focal sparse convolutional networks for 3d object detection, in: IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 5418–5427

    work page

  7. [6]

    Autoalign: Pixel-instance feature aggregation for multi-modal 3d object detection, in: International Joint Conference on Artificial Intelligence, pp

    Chen, Z., Li, Z., Zhang, S., Fang, L., Jiang, Q., Zhao, F., Zhou, B., Zhao, H., 2022c. Autoalign: Pixel-instance feature aggregation for multi-modal 3d object detection, in: International Joint Conference on Artificial Intelligence, pp. 827–833

    work page

  8. [7]

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

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

    work page 2025

Show all 59 references
  1. [8]

    Mmdetection3d: Openmmlab next-generation platform for general 3d object detection

    Contributors, M., 2020. Mmdetection3d: Openmmlab next-generation platform for general 3d object detection

    2020

  2. [9]

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

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

    2023

  3. [10]

    3d-lanenet:end-to-end3dmultiplelanedetection,in:ProceedingsoftheIEEE/CVF International Conference on Computer Vision, pp

    Garnett,N.,Cohen,R.,Pe’er,T.,Lahav,R.,Levi,D.,2019. 3d-lanenet:end-to-end3dmultiplelanedetection,in:ProceedingsoftheIEEE/CVF International Conference on Computer Vision, pp. 2921–2930

    2019

  4. [11]

    Bird’s-eye-view panoptic segmentation using monocular frontal view images

    Gosala, N., Valada, A., 2022. Bird’s-eye-view panoptic segmentation using monocular frontal view images. IEEE Robotics and Automation Letters 7, 1968–1975

    2022

  5. [12]

    Mapdistill:Boostingefficientcamera-based hd map construction via camera-lidar fusion model distillation, in: European Conference on Computer Vision, Springer

    Hao,X.,Li,R.,Zhang,H.,Li,D.,Yin,R.,Jung,S.,Park,S.I.,Yoo,B.,Zhao,H.,Zhang,J.,2024a. Mapdistill:Boostingefficientcamera-based hd map construction via camera-lidar fusion model distillation, in: European Conference on Computer Vision, Springer. pp. 166–183

  6. [13]

    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., Liu, Y., 2025. Msc-bench: Benchmarking and analyzing multi-sensor corruption for driving perception. arXiv preprint arXiv:2501.01037

    2025 arXiv

  7. [14]

    Is your hd map constructor reliable under sensor corruptions?

    Hao, X., Wei, M., Yang, Y., Zhao, H., Zhang, H., Zhou, Y., Wang, Q., Li, W., Kong, L., Zhang, J., 2024b. Is your hd map constructor reliable under sensor corruptions?

  8. [15]

    Mbfusion: A new multi-modal bev feature fusion method for hd map construction, in: 2024 IEEE International Conference on Robotics and Automation (ICRA), IEEE

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

    2024

  9. [16]

    Mixgen:Anewmulti-modaldataaugmentation,in:Proceedings of the IEEE/CVF winter conference on applications of computer vision, pp

    Hao,X.,Zhu,Y.,Appalaraju,S.,Zhang,A.,Zhang,W.,Li,B.,Li,M.,2023. Mixgen:Anewmulti-modaldataaugmentation,in:Proceedings of the IEEE/CVF winter conference on applications of computer vision, pp. 379–389

    2023

  10. [17]

    Deep residual learning for image recognition, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    He, K., Zhang, X., Ren, S., Sun, J., 2016a. Deep residual learning for image recognition, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

  11. [18]

    Deepresiduallearningforimagerecognition,in:IEEEConferenceonComputerVisionandPattern Recognition, pp

    He,K.,Zhang,X.,Ren,S.,Sun,J.,2016b. Deepresiduallearningforimagerecognition,in:IEEEConferenceonComputerVisionandPattern Recognition, pp. 770–778

  12. [19]

    Squeeze-and-excitation networks, in: Proceedings of the IEEE conference on computer vision and pattern recognition, pp

    Hu, J., Shen, L., Sun, G., 2018. Squeeze-and-excitation networks, in: Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 7132–7141

    2018

  13. [20]

    Pointpillars: Fastencoders forobject detectionfrom pointclouds, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp

    Lang,A.H., Vora,S., Caesar,H., Zhou,L., Yang,J., Beijbom,O., 2019. Pointpillars: Fastencoders forobject detectionfrom pointclouds, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 12697–12705

    2019

  14. [21]

    Hdmapnet: An online hd map construction and evaluation framework, in: IEEE International Conference on Robotics and Automation, pp

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

  15. [22]

    Bevformer: Learning bird’s-eye-view representation from multi- camera images via spatiotemporal transformers, in: European Conference on Computer Vision, pp

    Li, Z., Wang, W., Li, H., Xie, E., Sima, C., Lu, T., Qiao, Y., Dai, J., 2022b. Bevformer: Learning bird’s-eye-view representation from multi- camera images via spatiotemporal transformers, in: European Conference on Computer Vision, pp. 1–18

  16. [23]

    Deep continuous fusion for multi-sensor 3d object detection, in: European Conference on Computer Vision, pp

    Liang, M., Yang, B., Wang, S., Urtasun, R., 2018. Deep continuous fusion for multi-sensor 3d object detection, in: European Conference on Computer Vision, pp. 663–678

    2018

  17. [24]

    Bevfusion:Asimpleandrobustlidar-camerafusion framework, pp

    Liang,T.,Xie,H.,Yu,K.,Xia,Z.,Lin,Z.,Wang,Y.,Tang,T.,Wang,B.,Tang,Z.,2022. Bevfusion:Asimpleandrobustlidar-camerafusion framework, pp. 10421–10434

    2022

  18. [26]

    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., Wang, X., 2023b. Maptrv2: An end-to-end framework for online vectorized HD map construction. arXiv preprint arXiv:2308.05736

    arXiv

  19. [27]

    Feature pyramid networks for object detection, in: Proceedings of the IEEE conference on computer vision and pattern recognition, pp

    Lin, T.Y., Dollár, P., Girshick, R., He, K., Hariharan, B., Belongie, S., 2017. Feature pyramid networks for object detection, in: Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 2117–2125

    2017

  20. [28]

    Vectormapnet: End-to-end vectorized hd map learning, in: International Conference on Machine Learning, pp

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

  21. [29]

    Swintransformerv2:Scaling up capacity and resolution, in: International Conference on Computer Vision and Pattern Recognition, pp

    Liu,Z.,Hu,H.,Lin,Y.,Yao,Z.,Xie,Z.,Wei,Y.,Ning,J.,Cao,Y.,Zhang,Z.,Dong,L.,Wei,F.,Guo,B.,2022. Swintransformerv2:Scaling up capacity and resolution, in: International Conference on Computer Vision and Pattern Recognition, pp. 11999–12009

    2022

  22. [30]

    Swin transformer: Hierarchical vision transformer using shifted windows, in: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp

    Liu, Z., Lin, Y., Cao, Y., Hu, H., Wei, Y., Zhang, Z., Lin, S., Guo, B., 2021. Swin transformer: Hierarchical vision transformer using shifted windows, in: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 9992–10002

    2021

  23. [31]

    Bevfusion:Multi-taskmulti-sensorfusionwithunifiedbird’s-eye view representation, in: ICRA, IEEE

    Liu,Z.,Tang,H.,Amini,A.,Yang,X.,Mao,H.,Rus,D.L.,Han,S.,2023b. Bevfusion:Multi-taskmulti-sensorfusionwithunifiedbird’s-eye view representation, in: ICRA, IEEE. pp. 2774–2781

  24. [32]

    Visualizing data using t-SNE

    van der Maaten, L., Hinton, G., 2008. Visualizing data using t-SNE. Journal of Machine Learning Research . Hao et al.:Preprint submitted to Information Fusion Page 18 of 19 MapFusion: A Novel BEV Feature Fusion Network for Multi-modal Map Construction

    2008

  25. [33]

    Malkauthekar,M.,2013.Analysisofeuclideandistanceandmanhattandistancemeasureinfacerecognition,in:ThirdInternationalConference on Computational Intelligence and Information Technology (CIIT 2013), pp. 503–507

    2013

  26. [34]

    Calibrating deep neural networks using focal loss

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

    2020

  27. [35]

    Cross-view semantic segmentation for sensing surroundings

    Pan, B., Sun, J., Leung, H.Y.T., Andonian, A., Zhou, B., 2020. Cross-view semantic segmentation for sensing surroundings. IEEE Robotics and Automation Letters 5, 4867–4873

    2020

  28. [36]

    Lift, splat, shoot: Encoding images from arbitrary camera rigs by implicitly unprojecting to 3d, in: European Conference on Computer Vision, pp

    Philion, J., Fidler, S., 2020a. Lift, splat, shoot: Encoding images from arbitrary camera rigs by implicitly unprojecting to 3d, in: European Conference on Computer Vision, pp. 194–210

  29. [37]

    Lift, splat, shoot: Encoding images from arbitrary camera rigs by implicitly unprojecting to 3d, in: European Conference on Computer Vision, pp

    Philion, J., Fidler, S., 2020b. Lift, splat, shoot: Encoding images from arbitrary camera rigs by implicitly unprojecting to 3d, in: European Conference on Computer Vision, pp. 194–210

  30. [38]

    End-to-end vectorized hd-map construction with piecewise bézier curve, in: IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

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

  31. [39]

    End-to-end vectorized hd-map construction with piecewise bezier curve, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

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

  32. [40]

    Predicting semantic map representations from images using pyramid occupancy networks, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp

    Roddick, T., Cipolla, R., 2020a. Predicting semantic map representations from images using pyramid occupancy networks, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 11135–11144

  33. [41]

    Predicting semantic map representations from images using pyramid occupancy networks, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Roddick, T., Cipolla, R., 2020b. Predicting semantic map representations from images using pyramid occupancy networks, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 11138–11147

  34. [42]

    Orthographic feature transform for monocular 3d object detection

    Roddick, T., Kendall, A., Cipolla, R., 2018. Orthographic feature transform for monocular 3d object detection. arXiv preprint arXiv:1811.08188

    2018 arXiv

  35. [43]

    Attention is all you need, in: Advances in Neural Information Processing Systems, pp

    Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A.N., Kaiser, L., Polosukhin, I., 2017. Attention is all you need, in: Advances in Neural Information Processing Systems, pp. 5998–6008

    2017

  36. [44]

    Pointpainting: Sequential fusion for 3d object detection, in: IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Vora, S., Lang, A.H., Helou, B., Beijbom, O., 2020. Pointpainting: Sequential fusion for 3d object detection, in: IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 4603–4611

    2020

  37. [45]

    Pointaugmenting: Cross-modal augmentation for 3d object detection, in: IEEE Conference on Computer Vision and Pattern Recognition, pp

    Wang, C., Ma, C., Zhu, M., Yang, X., 2021. Pointaugmenting: Cross-modal augmentation for 3d object detection, in: IEEE Conference on Computer Vision and Pattern Recognition, pp. 11794–11803

    2021

  38. [46]

    Internimage: Exploring large-scale vision foundation models with deformable convolutions

    Wang, W., Dai, J., Chen, Z., Huang, Z., Li, Z., Zhu, X., Hu, X., Lu, T., Lu, L., Li, H., et al., 2022. Internimage: Exploring large-scale vision foundation models with deformable convolutions. arXiv preprint arXiv:2211.05778

    2022 arXiv

  39. [47]

    Argoverse 2: Next generation datasets for self-driving perception and forecasting, in: Proceedings of the Neural Information Processing Systems Track on Datasets and Benchmarks

    Wilson, B., Qi, W., Agarwal, T., Lambert, J., Singh, J., Khandelwal, S., Pan, B., Kumar, R., Hartnett, A., Pontes, J.K., Ramanan, D., Carr, P., Hays, J., 2021. Argoverse 2: Next generation datasets for self-driving perception and forecasting, in: Proceedings of the Neural Info...

    2021

  40. [48]

    M2bev: Multi-camera joint 3d detection and segmentation with unified birds-eye view representation

    Xie, E., Yu, Z., Zhou, D., Philion, J., Anandkumar, A., Fidler, S., Luo, P., Álvarez, J.M., 2022. M2bev: Multi-camera joint 3d detection and segmentation with unified birds-eye view representation. arXiv preprint arXiv:2204.05088

    2022 arXiv

  41. [49]

    Neuralmappriorforautonomousdriving,in:ProceedingsoftheIEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Xiong,X.,Liu,Y.,Yuan,T.,Wang,Y.,Wang,Y.,Zhao,H.,2023. Neuralmappriorforautonomousdriving,in:ProceedingsoftheIEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 17535–17544

    2023

  42. [50]

    Fusionpainting:Multimodalfusionwithadaptiveattentionfor3dobjectdetection, in: IEEE International Intelligent Transportation Systems Conference, pp

    Xu,S.,Zhou,D.,Fang,J.,Yin,J.,Zhou,B.,Zhang,L.,2021. Fusionpainting:Multimodalfusionwithadaptiveattentionfor3dobjectdetection, in: IEEE International Intelligent Transportation Systems Conference, pp. 3047–3054

    2021

  43. [51]

    SECOND: sparsely embedded convolutional detection

    Yan, Y., Mao, Y., Li, B., 2018. SECOND: sparsely embedded convolutional detection. Sensors 18, 3337

    2018

  44. [52]

    Bevformer v2: Adapting modern image backbones to bird’s-eye-view recognition via perspective supervision

    Yang, C., Chen, Y., Tian, H., Tao, C., Zhu, X., Zhang, Z., Huang, G., Li, H., Qiao, Y., Lu, L., et al., 2022. Bevformer v2: Adapting modern image backbones to bird’s-eye-view recognition via perspective supervision. arXiv preprint arXiv:2211.10439

    2022 arXiv

  45. [53]

    Center-based 3d object detection and tracking, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp

    Yin, T., Zhou, X., Krahenbuhl, P., 2021a. Center-based 3d object detection and tracking, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 11784–11793

  46. [54]

    Multimodal virtual point 3d detection, in: Conference on Neural Information Processing Systems, pp

    Yin, T., Zhou, X., Krähenbühl, P., 2021b. Multimodal virtual point 3d detection, in: Conference on Neural Information Processing Systems, pp. 16494–16507

  47. [55]

    Streammapnet:Streamingmappingnetworkforvectorizedonlinehdmapconstruction, in: IEEE/CVF Winter Conference on Applications of Computer Vision, pp

    Yuan,T.,Liu,Y.,Wang,Y.,Wang,Y.,Zhao,H.,2024. Streammapnet:Streamingmappingnetworkforvectorizedonlinehdmapconstruction, in: IEEE/CVF Winter Conference on Applications of Computer Vision, pp. 7356–7365

    2024

  48. [56]

    Onlinemapvectorizationforautonomousdriving:Arasterization perspective

    Zhang,G.,Lin,J.,Wu,S.,Song,Y.,Luo,Z.,Xue,Y.,Lu,S.,Wang,Z.,2023. Onlinemapvectorizationforautonomousdriving:Arasterization perspective. arXiv preprint arXiv:2306.10502

    2023 arXiv

  49. [57]

    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., Lu, J., 2022. Beverse: Unified perception and prediction in birds-eye-view for vision-centric autonomous driving. arXiv preprint arXiv:2205.09743

    2022 arXiv

  50. [58]

    Cross-view transformers for real-time map-view semantic segmentation, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp

    Zhou, B., Krähenbühl, P., 2022. Cross-view transformers for real-time map-view semantic segmentation, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 13760–13769

    2022

  51. [59]

    Monocular 3d vehicle detection using uncalibrated traffic cameras through homography, in: 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp

    Zhu, M., Zhang, S., Zhong, Y., Lu, P., Peng, H., Lenneman, J., 2021. Monocular 3d vehicle detection using uncalibrated traffic cameras through homography, in: 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 3814–3821. Hao et al.:Preprint su...

    2021

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.