Pith. sign in

REVIEW 3 major objections 4 minor 49 references

3DLabelProp: Geometric-Driven Domain Generalization for LiDAR Semantic Segmentation in Autonomous Driving

T0 review · 3 major / 4 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read This paper claims that geometry alone—accumulating registered scans and propagating static labels—can outperform learned domain-generalization methods for LiDAR semantic segmentation.

desk verdict Useful benchmark extension with an overclaimed SOTA result: the final comparison omits the pseudo-dense control that would isolate 3DLabelProp's actual contribution. read the letter →

arxiv 2501.14605 v1 pith:NUA3XQ3A submitted 2025-01-24 cs.CV

Jules Sanchez , Jean-Emmanuel Deschaud , François Goulette This is my paper · ORCID
classification cs.CV
keywords LiDARsemanticsegmentationdomaingeneralizationpseudo-densepointcloudslabelpropagationsensorshiftautonomousdriving3Dcloudcross-datasetevaluation
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 claims that domain generalization for LiDAR semantic segmentation can be achieved geometrically rather than by learning domain-invariant representations. The proposed method, 3DLabelProp, registers each new scan into a pseudo-dense cloud of previous scans, propagates labels of static objects to new points by proximity-weighted voting, and sends only the remaining dynamic or novel points to a deep network. Across seven real-world datasets and two training sources, the method reports higher cross-dataset mIoU than single-scan baselines and previous domain-generalization methods, and it is the only approach that extracts useful segmentation from a solid-state forward-facing LiDAR when trained on a 64-beam rotating sensor. The price is speed: the pipeline is below real-time, and its advantage depends on the odometry that builds the reference clouds being accurate.

What carries the argument

The central object is the pseudo-dense point cloud: the current scan merged with the previous $N_s$ registered scans, which makes the local geometry of static scenes nearly identical across sensors. The mechanism is the 4D-neighbor hypothesis—that every new point on a static object has a previously labeled point within a small distance in the registered cloud—combined with a propagation step that assigns labels by confidence- and distance-weighted voting (Eq. 1) and a clustering step that isolates dynamic or novel regions for the deep network. The propagation is accelerated by voxelizing the reference cloud, and cluster densification adds contextual points from the reference cloud so KPConv only has to process small, dense neighborhoods.

What would settle it

A direct test would take a sequence with deliberately corrupted odometry (as in the paper's SLAM-perturbation experiment) and count, per frame, the fraction of static points whose nearest labeled reference neighbor is farther than $d_p = 0.30$ m; if that fraction exceeds a few percent on a solid-state sensor, the geometric advantage over single-scan models should disappear.

Watch

Extended reading notes

Core claim

The paper's central claim is that domain generalization for LiDAR semantic segmentation does not need a learned alignment or augmentation strategy: it can be achieved geometrically by exploiting the temporal structure of the sensor stream. The authors propose 3DLabelProp, which registers the current scan into a reference cloud built from the previous 20 scans (using CT-ICP odometry), then labels every new point whose best nearby vote is a static class, with a confidence-weighted vote $c_j=\sum_{p_i\in\mathcal{N}(p_j)} w(i,j)\,\mathbf{1}_{w(i,j)>0.5}$ where $w(i,j)=e^{-\|p_i-p_j\|^2/d_p^2} c_i$ and $d_p=0.30$ m. Points that cannot be labeled geometrically (dynamic objects and new samples) are clustered, densified with reference points, and sent to a KPConv network trained only on such clusters. The paper reports that this pipeline outperforms single-scan models and previous domain-generalization methods on seven datasets, and that it is the only method that extracts meaningful segmentation from a solid-state forward-facing LiDAR when trained on a 64-beam rotating sensor (mIoU 59.3 vs 22.2 for the best single-scan competitor).

Load-bearing premise

The load-bearing assumption is the 4D-neighbor hypothesis: every new point on a static object has a previously labeled point within the propagation radius (0.30 m) in the registered reference cloud, and that earlier label and confidence are correct.

Editorial extensions

If this is right

  • A model trained on SemanticKITTI keeps essentially full accuracy on a 32-beam version of the same sensor (mIoU 61.7 vs 61.9), showing that resolution change alone no longer causes a drop.
  • On the solid-state PandaFF sensor, 3DLabelProp reaches mIoU 59.3 versus 22.2 for the best single-scan baseline, so front-facing solid-state LiDAR data can be segmented without retraining.
  • Switching the training source to nuScenes still yields the best cross-sensor results and a source-to-source gain (71.5 vs 70.2 mIoU), so the benefit is not tied to a particular sensor resolution.
  • Reflectivity, a sensor-specific channel, hurts generalization; removing it during training improves cross-dataset mIoU consistently across six architectures.
  • The geometric pipeline preserves source-to-source accuracy within 0.4 mIoU of the best single-scan method while avoiding the memory failures that naive pseudo-dense processing hits on dense data.

Reading between the lines

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

  • Beyond the paper: if the 4D-neighbor hypothesis holds, the same propagation step could turn any single-scan segmentation model into a temporally aware one at inference time, without retraining, by feeding it only the residual clusters.
  • Beyond the paper: the pseudo-dense representation's near-insensitivity to resolution suggests it could serve as a common coordinate space for merging multiple annotated datasets, reducing label-set engineering.
  • Beyond the paper: because the paper identifies KPConv as the speed bottleneck, a faster dense-point-cloud backbone could make the geometric generalization pipeline real-time, which is the main obstacle to on-vehicle use.
  • Beyond the paper: one could test the method on a sensor type never seen in any training set and measure whether the mIoU gap to single-scan baselines grows with registration drift; the paper's SLAM-noise ablation suggests it will.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper proposes 3DLabelProp, a geometry-based domain generalization method for LiDAR semantic segmentation. The method forms pseudo-dense point clouds by registering the current scan with the previous Ns=20 scans using CT-ICP, propagates labels from previous predictions in static regions using a distance-weighted vote, clusters the remaining points, densifies the clusters with reference points, segments the clusters with KPConv, and fuses the results. The authors also present a domain generalization benchmark across seven datasets, compare with existing DG methods (C&L, LiDOG, DGLSS), and provide ablations of the geometric hyperparameters. The central claim is that 3DLabelProp is state-of-the-art, outperforming both naive and learning-based domain generalization methods.

Significance. The paper has genuine strengths: it ships public code, evaluates on an unusually broad set of real LiDAR datasets, includes a careful analysis of pseudo-dense point clouds, and demonstrates large gains on the challenging PandaFF solid-state sensor relative to single-scan baselines. If the central claim were supported, the paper would make a useful contribution by showing that geometric pseudo-dense processing is a viable alternative to learning-based DG. However, the headline SOTA claim is not supported by the paper's own numbers, because the final benchmark omits the naive pseudo-dense baselines that serve as the natural control for the method and, in several cases, those baselines outperform 3DLabelProp.

major comments (3)
  1. [V-D / Table VIII vs Table II / Abstract] The abstract states that 3DLabelProp 'outperforms both naive and other domain generalization methods,' but this is contradicted by the paper's own tables. Section V-D (Table VIII) compares 3DLabelProp only against single-scan or sequence models and omits the pseudo-dense baselines introduced in Section IV-B (Table II). Under the same training source (SemanticKITTI) and label sets, Table II reports SRU-Net trained on naive pseudo-dense input reaching 61.7 mIoU on PandaFF, 49.8 on nuScenes, and 38.0 on ParisLuco3D, versus 59.3, 45.6, and 35.9 for 3DLabelProp in Table VIII; on SemanticPOSS the naive pseudo-dense SRU-Net is close (46.9 vs 47.2). Therefore the Section V-D claim that 3DLabelProp is 'the only method able to extract meaningful information from PandaFF' holds only among the single-scan baselines, and the 'outperforming naive' statement is not supported. The final benchmark must include pseudo-dense controls, and the claims must be qualified accordingly.
  2. [VI-A / Table XII] The comparison with Complete & Label uses C&L's unsupervised domain adaptation (UDA) results rather than its domain generalization results. UDA gives C&L access to unlabeled target-domain scans, so this is not a valid DG comparison; the authors acknowledge the issue in the text ('not entirely fair') but still present the numbers as evidence of effectiveness. The paper should either use C&L's DG numbers, even if limited to two classes, or explicitly reframe the table as a separate sanity check and avoid drawing DG conclusions from it.
  3. [IV-C, IV-D, and Table IV] The method's load-bearing assumption is the 4D-neighbor hypothesis, and its propagation step uses previous predictions at inference, whereas Section IV-D trains KPConv on clusters whose propagated labels come from ground-truth past inferences. This train-inference mismatch is not analyzed. Moreover, Table IV shows that adding SLAM noise to pseudo-dense input degrades naive SRU-Net by about 11 mIoU on SemanticKITTI (56.7 to 45.1) and by more on lower-resolution sensors; because 3DLabelProp's propagation depends on both registration quality and previous prediction quality, the paper should include an analogous perturbation study for 3DLabelProp and report how propagation errors accumulate over the Ns=20 scan window. Without this, the robustness of the central mechanism is not established.
minor comments (4)
  1. [V-A] The description of SemanticPOSS is repeated in two consecutive paragraphs with nearly identical wording; please merge them.
  2. [III-A / Table VIII / Figure 11 caption] There are several typos, including 'from from variations' in Section III-A, 'V oxel' in Table VIII, and 'Ground T ruth' in the Figure 11 caption; a proofreading pass is needed.
  3. [VIII / Table XVII] The inference-speed numbers in Table XVII should state whether they include the CT-ICP registration and label-propagation stages or only the neural-network forward pass; this is essential for interpreting the real-time discussion.
  4. [V-D / Table VIII] The N/A entry for Helix4D on Waymo in Table VIII is not explained; the table caption should note the reason, as is done for the N/A entries in Table II.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: 3DLabelProp is evaluated on external held-out datasets and its core prediction is a geometric propagation algorithm, not a refitted version of its inputs.

full rationale

3DLabelProp's claimed contribution is an algorithmic pipeline: accumulate registered scans with CT-ICP, propagate static labels from the reference cloud via Eq. 1, cluster and densify residual points, and run a source-trained KPConv on the clusters. None of these steps is fitted to the target benchmarks; the hyperparameters dp=0.30 m, Kc=20, and Ns=20 are fixed and ablated on NS->SK, and mIoU is computed against held-out target ground truth. The propagation equation is explicitly a nearest-neighbor vote over previous predictions, so it is a self-referential recursion at inference time, but the paper does not present it as a first-principles derivation nor as a prediction whose output equals an input: the evaluation labels are external. Self-citations ([5] for the original formulation, [43] for the ParisLuco3D dataset, and [49] for CT-ICP registration) point to published, code-released artifacts and are not invoked as an authority to forbid alternatives or to supply a uniqueness result. The main caveats are not circularity: Section VIII admits the method is below real-time, and Section V-D's Table VIII omits the naive pseudo-dense baselines from Table II, so the abstract's 'outperforming both naive and other domain generalization methods' is under-supported as a comparison claim; similarly, Section VI-A notes the C&L comparison is 'not entirely fair due to the differing backbones.' These are benchmark-completeness and comparison-fairness issues, not input-output equivalences. The central result remains externally grounded in previously unseen target datasets, so the derivation chain does not reduce to its own inputs.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The central claim rests on hyperparameters (dp, Kc, Ns, voxel sizes) and on domain assumptions about SLAM accuracy, temporal sampling density, static/dynamic taxonomy, and label-set semantics. The paper provides one ablation for the main hyperparameters but no multi-seed statistics, and the SLAM assumption is tested only by injecting artificial noise.

free parameters (6)
  • dp, label propagation distance scale = 0.30 m (default)
    Controls the weight of neighboring points in label propagation. Chosen via ablation on NS to SK in Table XV, then used across all experiments.
  • Kc, number of clusters = 20 (default)
    Number of K-means clusters for residual dynamic points. Selected via ablation; Kc=5 caused memory issues on SemanticKITTI.
  • Ns, number of past scans = 20 (default)
    Number of previous scans accumulated into the pseudo-dense cloud. Selected via ablation as a balance between context and trail noise.
  • Propagation voxel size = 0.80 m
    Voxelization used for neighborhood extraction in label propagation. Fixed in all experiments, not ablated.
  • Cluster densification voxel size = 2 m with 3x3x3 sub-voxels
    Voxel size for adding context points from the reference cloud to clusters. Fixed in all experiments, not ablated.
  • Reference cloud sub-sampling and range cutoff = 5 cm grid, 75 m range
    Engineering choices to reduce memory and processing time for the reference cloud.
assumptions (5)
  • domain assumption CT-ICP SLAM registration is accurate enough to align scans into a common reference frame across all seven datasets, including under large sensor shifts.
    The pseudo-dense point clouds are built with CT-ICP and registration is not treated as a learned component. Section IV-B and Table IV show performance degrades when SLAM is perturbed.
  • domain assumption 4D-neighbor hypothesis: every newly sampled static point has a previous 4D neighbor in the reference cloud that samples the same object within the propagation radius.
    This is the foundation of the label propagation step, stated in Section IV-C: static objects are assumed to have been previously sampled.
  • domain assumption The static/dynamic label split is known and fixed for each dataset, and dynamic labels such as vehicles and pedestrians are never propagated.
    Section IV-C.1 divides the label set into static and dynamic subsets; errors in this taxonomy would propagate wrong labels or send static points to the network unnecessarily.
  • domain assumption At inference time, previous predictions used for propagation are sufficiently reliable to label static points in the current scan.
    Training uses ground-truth propagated labels, while inference uses past inferences. This train/test mismatch is stated in Section IV-D and is not analyzed in the paper.
  • domain assumption Intersection label sets preserve semantic correspondence across datasets without label shift.
    The evaluation protocol in Section III-C constructs label sets as intersections of dataset labels; this assumes shared class meanings and ignores annotation inconsistencies beyond label naming.

how reviews work

0 comments
Cite this review

Pith. "Pith review of 3DLabelProp: Geometric-Driven Domain Generalization for LiDAR Semantic Segmentation in Autonomous Driving." pith.science (2026). https://pith.science/paper/NUA3XQ3A

@misc{pith2026250114605,
  author       = {Pith},
  title        = {Pith review of: 3DLabelProp: Geometric-Driven Domain Generalization for LiDAR Semantic Segmentation in Autonomous Driving},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NUA3XQ3A}},
  note         = {Machine review of arXiv:2501.14605}
}
read the original abstract

Domain generalization aims to find ways for deep learning models to maintain their performance despite significant domain shifts between training and inference datasets. This is particularly important for models that need to be robust or are costly to train. LiDAR perception in autonomous driving is impacted by both of these concerns, leading to the emergence of various approaches. This work addresses the challenge by proposing a geometry-based approach, leveraging the sequential structure of LiDAR sensors, which sets it apart from the learning-based methods commonly found in the literature. The proposed method, called 3DLabelProp, is applied on the task of LiDAR Semantic Segmentation (LSS). Through extensive experimentation on seven datasets, it is demonstrated to be a state-of-the-art approach, outperforming both naive and other domain generalization methods.

Figures

Figures reproduced from arXiv: 2501.14605 by the authors.

Figure 1
Figure 1. Illustration of our approach using pseudo-dense points for domain generalization of LiDAR semantic Segmentation in autonomous driving (the blue [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 3
Figure 3. Illustration of scene shift: On the left, a scan from the SemanticKITTI [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 4
Figure 4. Illustration of sensor shift: Both scans were acquired simultaneously [PITH_FULL_IMAGE:figures/full_fig_p003_4.png] view at source ↗
Figures from the paper (7 more)
Figure 5
Figure 5. Figure 5: Illustration of the trail phenomenon. On the left, a section of a point [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Overview of the 3DLabelProp method. Points from the current scan are accumulated with points from previous scans. A geometric propagation labels [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Propagation module of our 3DLabelProp method. We use the [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Results of the propagation module. In black, the points considered [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]
Figure 9
Figure 9. Figure 9: Cluster densification module. We subdivide the voxels occupied by [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 10
Figure 10. Figure 10: On the left, an extracted cluster from the residual points in the [PITH_FULL_IMAGE:figures/full_fig_p008_10.png]
Figure 11
Figure 11. Figure 11: Qualitative results for KPConv [10], SPVCNN [17], and 3DLabelProp trained on SemanticKITTI and tested on Panda64 (top row) and PandaFF [PITH_FULL_IMAGE:figures/full_fig_p011_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references · 43 canonical work pages

  1. [1]

    Single domain generalization for lidar semantic segmentation,

    H. Kim, Y . Kang, C. Oh, and K.-J. Yoon, “Single domain generalization for lidar semantic segmentation,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , 2023, pp. 17 587–17 598

    work page 2023

  2. [2]

    Walking your lidog: A journey through multiple domains for lidar semantic segmentation,

    C. Saltori, A. O ˇsep, E. Ricci, and L. Leal-Taix´e, “Walking your lidog: A journey through multiple domains for lidar semantic segmentation,” in IEEE/CVF International Conference on Computer Vision (ICCV) , 2023, pp. 196–206

    work page 2023

  3. [3]

    3d-vfield: Adversarial augmentation of point clouds for domain generalization in 3d object detection,

    A. Lehner, S. Gasperini, A. Marcos-Ramiro, M. Schmidt, M.-A. N. Mahani, N. Navab, B. Busam, and F. Tombari, “3d-vfield: Adversarial augmentation of point clouds for domain generalization in 3d object detection,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), June 2022, pp. 17 295–17 304

    work page 2022

  4. [4]

    Complete & label: A domain adaptation approach to semantic segmentation of lidar point clouds,

    L. Yi, B. Gong, and T. Funkhouser, “Complete & label: A domain adaptation approach to semantic segmentation of lidar point clouds,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021, pp. 15 358–15 368

    work page 2021

  5. [5]

    Domain generalization of 3d semantic segmentation in autonomous driving,

    J. Sanchez, J. Deschaud, and F. Goulette, “Domain generalization of 3d semantic segmentation in autonomous driving,” in IEEE/CVF International Conference on Computer Vision (ICCV), 2023, pp. 18 031– 18 041. 14

    work page 2023

  6. [6]

    Pointnet: Deep learning on point sets for 3d classification and segmentation,

    C. R. Qi, H. Su, K. Mo, and L. J. Guibas, “Pointnet: Deep learning on point sets for 3d classification and segmentation,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2017, pp. 652–660

    work page 2017

  7. [7]

    Pointnet++: deep hierarchical feature learning on point sets in a metric space,

    C. R. Qi, L. Yi, H. Su, and L. J. Guibas, “Pointnet++: deep hierarchical feature learning on point sets in a metric space,” in Proceedings of the 31st International Conference on Neural Information Processing Systems, ser. NIPS’17. Red Hook, NY , USA: Curran Associates Inc., 2017, p. 5105–5114

    2017

  8. [8]

    Randla-net: Efficient semantic segmentation of large- scale point clouds,

    Q. Hu, B. Yang, L. Xie, S. Rosa, Y . Guo, Z. Wang, N. Trigoni, and A. Markham, “Randla-net: Efficient semantic segmentation of large- scale point clouds,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020, pp. 11 108–11 117

    work page 2020

Show all 49 references
  1. [9]

    Spidercnn: Deep learning on point sets with parameterized convolutional filters,

    Y . Xu, T. Fan, M. Xu, L. Zeng, and Y . Qiao, “Spidercnn: Deep learning on point sets with parameterized convolutional filters,” in Proceedings of the European conference on computer vision (ECCV), 2018, pp. 87–102

    2018

  2. [10]

    Kpconv: Flexible and deformable convolution for point clouds,

    H. Thomas, C. R. Qi, J.-E. Deschaud, B. Marcotegui, F. Goulette, and L. J. Guibas, “Kpconv: Flexible and deformable convolution for point clouds,” in IEEE/CVF International Conference on Computer Vision (ICCV), 2019, pp. 6411–6420

    2019

  3. [11]

    Rangenet++: Fast and accurate lidar semantic segmentation,

    A. Milioto, I. Vizzo, J. Behley, and C. Stachniss, “Rangenet++: Fast and accurate lidar semantic segmentation,” in IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , 2019, pp. 4213– 4220

    2019

  4. [12]

    Squeezesegv3: Spatially-adaptive convolution for effi- cient point-cloud segmentation,

    C. Xu, B. Wu, Z. Wang, W. Zhan, P. Vajda, K. Keutzer, and M. Tomizuka, “Squeezesegv3: Spatially-adaptive convolution for effi- cient point-cloud segmentation,” in Proceedings of the European Con- ference on Computer Vision (ECCV) , 2020, pp. 1–19

    2020

  5. [13]

    Cenet: Toward concise and efficient lidar semantic segmentation for autonomous driving,

    H.-X. Cheng, X.-F. Han, and G.-Q. Xiao, “Cenet: Toward concise and efficient lidar semantic segmentation for autonomous driving,” in IEEE International Conference on Multimedia and Expo (ICME) , 2022, pp. 01–06

    2022

  6. [14]

    Polarnet: An improved grid representation for online lidar point clouds semantic segmentation,

    Y . Zhang, Z. Zhou, P. David, X. Yue, Z. Xi, B. Gong, and H. Foroosh, “Polarnet: An improved grid representation for online lidar point clouds semantic segmentation,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , 2020, pp. 9601–9610

    2020

  7. [15]

    4d spatio-temporal convnets: Minkowski convolutional neural networks,

    C. Choy, J. Gwak, and S. Savarese, “4d spatio-temporal convnets: Minkowski convolutional neural networks,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , 2019, pp. 3075– 3084

    2019

  8. [16]

    Cylindrical and asymmetrical 3d convolution networks for lidar segmentation,

    X. Zhu, H. Zhou, T. Wang, F. Hong, Y . Ma, W. Li, H. Li, and D. Lin, “Cylindrical and asymmetrical 3d convolution networks for lidar segmentation,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021, pp. 9934–9943

    2021

  9. [17]

    Search- ing efficient 3d architectures with sparse point-voxel convolution,

    H. Tang, Z. Liu, S. Zhao, Y . Lin, J. Lin, H. Wang, and S. Han, “Search- ing efficient 3d architectures with sparse point-voxel convolution,” in Proceedings of the European Conference on Computer Vision (ECCV) , 2020, pp. 685–702

    2020

  10. [18]

    Asap-net: Attention and structure aware point cloud sequence segmentation,

    H. Cao, Y . Lu, C. Lu, B. Pang, G. Liu, and A. Yuille, “Asap-net: Attention and structure aware point cloud sequence segmentation,” arXiv preprint arXiv:2008.05149, 2020

    2008 arXiv

  11. [19]

    Spsequencenet: Semantic segmentation network on 4d point clouds,

    H. Shi, G. Lin, H. Wang, T.-Y . Hung, and Z. Wang, “Spsequencenet: Semantic segmentation network on 4d point clouds,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020, pp. 4574–4583

    2020

  12. [20]

    Lidar-based recurrent 3d semantic segmentation with temporal memory alignment,

    F. Duerr, M. Pfaller, H. Weigel, and J. Beyerer, “Lidar-based recurrent 3d semantic segmentation with temporal memory alignment,” in Inter- national Conference on 3D Vision (3DV) , 2020, pp. 781–790

    2020

  13. [21]

    Meteornet: Deep learning on dynamic 3d point cloud sequences,

    X. Liu, M. Yan, and J. Bohg, “Meteornet: Deep learning on dynamic 3d point cloud sequences,” in IEEE/CVF International Conference on Computer Vision (ICCV) , 2019, pp. 9245–9254

    2019

  14. [22]

    Pstnet: Point spatio-temporal convolution on point cloud sequences,

    H. Fan, X. Yu, Y . Ding, Y . Yang, and M. Kankanhalli, “Pstnet: Point spatio-temporal convolution on point cloud sequences,” in International Conference on Learning Representations (ICLR) , 2021

    2021

  15. [23]

    Online segmentation of lidar sequences: Dataset and algorithm,

    R. Loiseau, M. Aubry, and L. Landrieu, “Online segmentation of lidar sequences: Dataset and algorithm,” in Proceedings of the European Conference on Computer Vision (ECCV) , 2022, pp. 301–317

    2022

  16. [24]

    4d panoptic lidar segmentation,

    M. Aygun, A. Osep, M. Weber, M. Maximov, C. Stachniss, J. Behley, and L. Leal-Taixe, “4d panoptic lidar segmentation,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , June 2021, pp. 5527–5537

    2021

  17. [25]

    4d-stop: Panoptic segmentation of 4d lidar using spatio-temporal ob- ject proposal generation and aggregation,

    L. Kreuzberg, I. E. Zulfikar, S. Mahadevan, F. Engelmann, and B. Leibe, “4d-stop: Panoptic segmentation of 4d lidar using spatio-temporal ob- ject proposal generation and aggregation,” in European Conference on Computer Vision Workshop, 2022

    2022

  18. [26]

    Generalizing to unseen domains: A survey on domain generalization,

    J. Wang, C. Lan, C. Liu, Y . Ouyang, T. Qin, W. Lu, Y . Chen, W. Zeng, and P. Yu, “Generalizing to unseen domains: A survey on domain generalization,” IEEE Transactions on Knowledge and Data Engineering, pp. 1–1, 2022

    2022

  19. [27]

    Domain general- ization: A survey,

    K. Zhou, Z. Liu, Y . Qiao, T. Xiang, and C. C. Loy, “Domain general- ization: A survey,” IEEE Transactions on Pattern Analysis and Machine Intelligence, pp. 1–20, 2022

    2022

  20. [28]

    Model-agnostic meta-learning for fast adaptation of deep networks,

    C. Finn, P. Abbeel, and S. Levine, “Model-agnostic meta-learning for fast adaptation of deep networks,” in International conference on machine learning. PMLR, 2017, pp. 1126–1135

    2017

  21. [29]

    Domain generalization by solving jigsaw puzzles,

    F. M. Carlucci, A. D’Innocente, S. Bucci, B. Caputo, and T. Tommasi, “Domain generalization by solving jigsaw puzzles,” in IEEE/CVF Con- ference on Computer Vision and Pattern Recognition (CVPR), 2019, pp. 2229–2238

    2019

  22. [30]

    Generalizing to unseen domains via adversarial data augmentation,

    R. V olpi, H. Namkoong, O. Sener, J. C. Duchi, V . Murino, and S. Savarese, “Generalizing to unseen domains via adversarial data augmentation,” Advances in neural information processing systems , vol. 31, 2018

    2018

  23. [31]

    Two at once: Enhancing learning and generalization capacities via ibn-net,

    X. Pan, P. Luo, J. Shi, and X. Tang, “Two at once: Enhancing learning and generalization capacities via ibn-net,” in Proceedings of the Euro- pean Conference on Computer Vision (ECCV) , 2018, pp. 464–479

    2018

  24. [32]

    Domain generalization using a mixture of multiple latent domains,

    T. Matsuura and T. Harada, “Domain generalization using a mixture of multiple latent domains,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 34, 2020, pp. 11 749–11 756

    2020

  25. [33]

    3d adversarial augmentations for robust out-of-domain predictions,

    A. Lehner, S. Gasperini, A. Marcos-Ramiro, M. Schmidt, N. Navab, B. Busam, and F. Tombari, “3d adversarial augmentations for robust out-of-domain predictions,” International Journal of Computer Vision , pp. 1–33, 2023

    2023

  26. [34]

    Cola: Coarse-label multi- source lidar semantic segmentation for autonomous driving,

    J. Sanchez, J.-E. Deschaud, and F. Goulette, “Cola: Coarse-label multi- source lidar semantic segmentation for autonomous driving,” 2023

    2023

  27. [35]

    Mdt3d: Multi-dataset training for lidar 3d object detection generalization,

    L. Soum-Fontez, J.-E. Deschaud, and F. Goulette, “Mdt3d: Multi-dataset training for lidar 3d object detection generalization,” in IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , 2023, pp. 5765–5772

    2023

  28. [36]

    Benchmarking the robustness of lidar semantic segmentation models,

    X. Yan, C. Zheng, Z. Li, S. Cui, and D. Dai, “Benchmarking the robustness of lidar semantic segmentation models,” ArXiv, vol. abs/2301.00970, 2023

    2023 arXiv

  29. [37]

    Semantickitti: A dataset for semantic scene understanding of lidar sequences,

    J. Behley, M. Garbade, A. Milioto, J. Quenzel, S. Behnke, C. Stachniss, and J. Gall, “Semantickitti: A dataset for semantic scene understanding of lidar sequences,” in IEEE/CVF International Conference on Computer Vision (ICCV), 2019, pp. 9296–9306

    2019

  30. [38]

    Learning to generalize: Meta-learning for domain generalization,

    D. Li, Y . Yang, Y .-Z. Song, and T. Hospedales, “Learning to generalize: Meta-learning for domain generalization,” in Proceedings of the AAAI conference on artificial intelligence , vol. 32, 2018

    2018

  31. [39]

    nuscenes: A multimodal dataset for autonomous driving,

    H. Caesar, V . Bankiti, A. H. Lang, S. V ora, V . E. Liong, Q. Xu, A. Kr- ishnan, Y . Pan, G. Baldan, and O. Beijbom, “nuscenes: A multimodal dataset for autonomous driving,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , 2020, pp. 11 618–11 628

    2020

  32. [40]

    Semanticposs: A point cloud dataset with large quantity of dynamic instances,

    Y . Pan, B. Gao, J. Mei, S. Geng, C. Li, and H. Zhao, “Semanticposs: A point cloud dataset with large quantity of dynamic instances,” in IEEE Intelligent Vehicles Symposium (IV) , 2020, pp. 687–693

    2020

  33. [41]

    Pandaset: Advanced sensor suite dataset for autonomous driving,

    P. Xiao, Z. Shao, S. Hao, Z. Zhang, X. Chai, J. Jiao, Z. Li, J. Wu, K. Sun, K. Jiang et al. , “Pandaset: Advanced sensor suite dataset for autonomous driving,” in IEEE International Intelligent Transportation Systems Conference (ITSC) . IEEE, 2021, pp. 3095–3101

    2021

  34. [42]

    Scalability in perception for autonomous driving: Waymo open dataset,

    P. Sun, H. Kretzschmar, X. Dotiwalla, A. Chouard, V . Patnaik, P. Tsui, J. Guo, Y . Zhou, Y . Chai, B. Caine, V . Vasudevan, W. Han, J. Ngiam, H. Zhao, A. Timofeev, S. Ettinger, M. Krivokon, A. Gao, A. Joshi, Y . Zhang, J. Shlens, Z. Chen, and D. Anguelov, “Scalability in perc...

    2020

  35. [43]

    Paris- luco3d: A high-quality target dataset for domain generalization of lidar perception,

    J. Sanchez, L. Soum-Fontez, J.-E. Deschaud, and F. Goulette, “Paris- luco3d: A high-quality target dataset for domain generalization of lidar perception,” IEEE Robotics and Automation Letters , vol. 9, no. 6, pp. 5496–5503, 2024

    2024

  36. [44]

    Squeezeseg: Convolutional neural nets with recurrent crf for real-time road-object segmentation from 3d lidar point cloud,

    B. Wu, A. Wan, X. Yue, and K. Keutzer, “Squeezeseg: Convolutional neural nets with recurrent crf for real-time road-object segmentation from 3d lidar point cloud,” in IEEE International Conference on Robotics and Automation (ICRA) . IEEE, 2018, pp. 1887–1893

    2018

  37. [45]

    Carla: An open urban driving simulator,

    A. Dosovitskiy, G. Ros, F. Codevilla, A. Lopez, and V . Koltun, “Carla: An open urban driving simulator,” in Conference on robot learning . PMLR, 2017, pp. 1–16

    2017

  38. [46]

    Kitti-carla: a kitti-like dataset generated by carla simulator,

    J.-E. Deschaud, “Kitti-carla: a kitti-like dataset generated by carla simulator,” arXiv preprint arXiv:2109.00892 , 2021

    2021 arXiv

  39. [47]

    Transfer learning from synthetic to real lidar point cloud for semantic segmentation,

    A. Xiao, J. Huang, D. Guan, F. Zhan, and S. Lu, “Transfer learning from synthetic to real lidar point cloud for semantic segmentation,” in Proceedings of the AAAI Conference on Artificial Intelligence , vol. 36, 2022, pp. 2795–2803. 15

    2022

  40. [48]

    Compositional semantic mix for domain adaptation in point cloud segmentation,

    C. Saltori, F. Galasso, G. Fiameni, N. Sebe, F. Poiesi, and E. Ricci, “Compositional semantic mix for domain adaptation in point cloud segmentation,” IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023

    2023

  41. [49]

    Ct-icp: Real-time elastic lidar odometry with loop closure,

    P. Dellenbach, J.-E. Deschaud, B. Jacquet, and F. Goulette, “Ct-icp: Real-time elastic lidar odometry with loop closure,” in International Conference on Robotics and Automation (ICRA) , 2022, pp. 5580–5586. Jules Sanchez received his PhD degree in computer science and robotics...

    2022

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.