Pith. sign in

REVIEW 4 major objections 6 minor 35 references

3D Can Be Explored In 2D: Pseudo-Label Generation for LiDAR Point Clouds Using Sensor-Intensity-Based 2D Semantic Segmentation

T0 review · 4 major / 6 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 A 2D segmentation model trained on street images can label 3D LiDAR scenes when the point cloud is rendered as intensity-only views.

desk verdict Real but narrow novelty—intensity-rendered views instead of camera images—with a solid empirical core and one untested domain-transfer assumption. read the letter →

arxiv 2505.03300 v1 pith:SEAXRYDK submitted 2025-05-06 cs.CV

Andrew Caunes , Thierry Chateau , Vincent Frémont This is my paper · ORCID
classification cs.CV
keywords LiDARpointclouds3Dsemanticsegmentationpseudo-labelgeneration2D-to-3Dlabeltransfersensorintensityrenderingmulti-viewvotingunsuperviseddomainadaptationdrivingscenes
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper sets out to show that 3D LiDAR semantic segmentation can be obtained without any 3D annotation and without camera images at inference time. The trick is to align the LiDAR scans of a scene, color each point by its sensor intensity, render many virtual-camera 2D views, and run a 2D segmentation model trained on ordinary street photographs. Each view's labels are projected back onto the 3D points and merged by a voting rule to produce pseudo-labels. On the five static classes considered, the pseudo-labels reach a mean IoU of about 0.49 against ground truth, and training a 3D model on them for unsupervised domain adaptation performs on par with recent methods that use temporal information. If this holds, expensive 3D annotation could be replaced by 2D annotation for static scene classes.

What carries the argument

The load-bearing mechanism is multi-view rendering plus vote fusion. Each aligned LiDAR scene is converted into a large number of greyscale images by projecting points with only their sensor-intensity values into virtual cameras placed near the sensor trajectory. A pretrained 2D segmentation model then produces per-view class logits and hard masks, which are back-projected onto the 3D points and accumulated as votes. Three election estimators are compared: summing hard one-hot votes, summing softmax logits (summing soft votes), and multiplying logits across views (compounding soft votes). Summing soft votes performs best, and the ablation shows that the number of views dominates: dropping from 600 to 100 views cuts mean IoU by more than 86%. The RGB-to-greyscale image augmentation applied when training the 2D model is what makes the camera-trained model usable on intensity renderings.

What would settle it

Take the same aligned scenes and replace each point's real intensity with random values or a flat color, run the same 2D model and voting pipeline, and compare pseudo-label IoU to the reported roughly 0.49; if the score barely drops, the sensor intensity is not what drives the recognition, contradicting the paper's domain-transfer explanation.

Watch

Extended reading notes

Core claim

The central claim is that direct inference of a 2D segmentation model on intensity-only renderings of dense aligned point clouds is enough to produce useful 3D semantic labels. The paper demonstrates this by coloring each LiDAR point with its measured intensity, clipping and rescaling that intensity, aligning an entire scan sequence into one dense scene, sampling hundreds of virtual camera poses along the sensor trajectory, and running the 2D model on each view. Softmax logits and hard masks are projected back to 3D and accumulated; the sum-of-soft-votes estimator, which totals the logits across views and takes the argmax per point, gives the best results. On the five static classes the best average IoU is about 0.50, with a 0.81 IoU for driveable surface, while building-like manmade structures score worst at 0.37 due to occlusion and limited viewpoint coverage. In the domain-adaptation setting, a 3D model trained on these pseudo-labels reaches 54.2 mean IoU without source pretraining and 61.4 with pretraining, about 6.8 points below a 1% weakly supervised baseline and comparable to the strongest compared UDA method.

Load-bearing premise

The whole pipeline depends on a 2D segmentation model trained on greyscale camera photographs recognizing the same semantic classes in intensity-rendered LiDAR views without any dedicated cross-domain training; if that transfer fails, the method has no labels to offer.

Editorial extensions

If this is right

  • Static classes such as roads, sidewalks, terrain, and vegetation can be pseudo-labelled across entire LiDAR sequences using only 2D image annotations, removing the need to hand-label 3D point clouds for those classes.
  • At inference the pipeline needs only LiDAR scans and sensor poses; cameras can be absent, which widens deployment to platforms without camera rigs.
  • The generated pseudo-labels are strong enough to serve as target-domain supervision for unsupervised domain adaptation, reaching results comparable to a temporal UDA method without using temporal consistency.
  • Performance scales strongly with the number and diversity of rendered views, so the method's quality is tied to ray-casting effort rather than to 3D network capacity.
  • The method is explicitly scoped to static classes; moving objects are distorted by scan alignment and are left to future work.

Reading between the lines

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

  • The paper does not test this, but the weak classes (buildings, vegetation) appear limited by occlusion and viewpoint coverage rather than by 2D recognition quality, so adding depth-aware occlusion masking during back-projection could raise pseudo-label IoU without retraining the 2D model.
  • Because the RGB-to-greyscale augmentation is already a crude domain transfer, training the 2D model directly on synthetic intensity renderings, or fine-tuning it on a few rendered frames, is a natural extension that could close the remaining domain gap; the paper does not test this.
  • The vote counts themselves are a per-point confidence signal, so filtering pseudo-labels by agreement across views before UDA training is an untested lever that could improve the 3D model over unfiltered training.
  • If 2D segmentation models continue to improve, this pipeline should inherit those gains for 3D classes, making the zero-3D-annotation route progressively more attractive without any 3D-specific changes.
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

4 major / 6 minor

Summary. The paper proposes a pipeline for generating pseudo-labels for LiDAR point cloud semantic segmentation without 3D annotations or camera images at inference time. Point clouds from a sequence of scans are aligned and colored only by sensor intensity; many virtual views are rendered from randomized poses; a 2D segmentation model trained on greyscale-converted Mapillary Vistas images is applied to each view; the resulting labels and logits are back-projected and fused by voting. The pseudo-labels are evaluated against nuScenes ground truth on five static classes, and their usefulness is tested in an unsupervised domain adaptation setting (SemanticKITTI to nuScenes), where the method reaches 61.4 mIoU with pretraining, comparable to T-UDA.

Significance. If the central premise holds, the method is a meaningful contribution: it is the first, to the authors' knowledge, to propagate 2D labels to generated LiDAR-intensity views, removing the need for posed camera images at inference. The paper provides a useful ablation study, evaluates on a large dataset, and is honest about limitations. However, the load-bearing domain-transfer assumption is never measured in isolation, the experiments are single-run without error bars, and the comparison to T-UDA is complicated by a large naive-baseline discrepancy. These issues make the quantitative claims less definitive than the text suggests.

major comments (4)
  1. [Section III / IV-A1 / Conclusion] The domain-transfer assumption that a 2D model trained on greyscale-converted Mapillary Vistas images can segment rendered LiDAR-intensity views is never measured directly. Every reported quality metric in Fig. 2 and Table I is computed after multi-view voting and 3D back-projection, so the per-view accuracy of the 2D model is confounded with the fusion estimator. The Conclusion's own admission that 'the limited domain in which the views can be generated without increasing the domain shift with the 2D dataset' remains an unresolved challenge is itself evidence that the key premise needs separate validation. Please add a direct evaluation of the 2D model on individual rendered views before fusion, for example per-view IoU on a held-out subset of generated views, and report the numbers per class.
  2. [Table I / Section IV-A2] All ablation results come from a single run with no error bars, multiple seeds, or significance tests. Several pairwise differences in Table I, such as (d) at 49.95 mIoU versus (e) at 49.77, and (f) at 53.14 versus (g) at 53.12, are small enough to be plausibly within run-to-run variance, yet the text draws conclusions from them, including the preference for 'summing soft votes' over 'compounding soft votes'. Please provide confidence intervals or repeated-seed results and base the claims about estimator choice and randomization levels on those results.
  3. [Table II / Section IV-B2] The comparison with T-UDA [6] is not apples-to-apples because the authors report a Naive no-DA mIoU of 7.0 while [6] reports 34.18 on the same classes, a discrepancy the paper attributes to pretraining augmentations without evidence. If the source-pretrained model used here is not trained under the same protocol as in [6], the UDA improvements in Table II are not directly comparable to the published numbers. Please reproduce the pretraining protocol of [6], or report the [6] Naive baseline alongside the authors' own Naive baseline in the same table, and temper the statement that the method is 'on-par' with T-UDA accordingly.
  4. [Fig. 2 / Section IV-A / Abstract] The headline pseudo-label mIoU is computed on a cropped point set (points beyond 30 m on each side and 10 m in height are removed) and after merging sidewalk and terrain, while the abstract and introduction claim '3D semantic segmentation' of whole scenes. These evaluation choices are post-hoc and are not reflected in the stated mIoU. Please report uncropped and unmerged metrics, or explicitly state the restricted evaluation domain in the abstract and conclusion so readers can judge the generality of the claim.
minor comments (6)
  1. [Fig. 2 / Section IV-A2] The figure reports an average IoU of 0.78 for the road class, while the text states an average IoU of 0.81 for the same class; please harmonize the numbers.
  2. [Algorithm 1] The pose-sampling algorithm uses the interval '1 < i < M', which should presumably be '1 <= i <= M', and the number of sampled poses K is not otherwise constrained by M; please clarify the intended range.
  3. [Table I] The header row of Table I is ambiguous because the column labels are displayed together with check-mark rows rather than as clearly separated column names; please render the table with explicit column headers.
  4. [Table II] The '1h' baseline row in Table II is not defined in the text; please specify how many annotated scans it uses and how the model is trained.
  5. [Section III] The compounding estimator multiplies raw logits from each view, but no normalization is described before the product; the text mentions a possible missing prior normalization, but please state explicitly whether softmax-normalized probabilities or raw logits were compounded.
  6. [Abstract] The abstract contains wording errors such as 'not other modality for inference'; please revise the language throughout for clarity and grammatical correctness.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: all load-bearing supervision comes from external datasets and the evaluation is against independent ground truth.

full rationale

The paper's derivation chain is self-contained with respect to circularity. The 2D segmentation model is trained on the Mapillary Vistas dataset, which is external to the authors' pipeline and to the nuScenes benchmark used for evaluation. The generated pseudo-labels are produced by rendering LiDAR intensity views, running the externally trained 2D model on those views, and back-projecting the predictions with a voting estimator. The central claim is then evaluated by comparing these pseudo-labels to nuScenes ground-truth annotations (Fig. 2) and by using them to train 3D models for unsupervised domain adaptation (Table II). There is no equation in which a predicted quantity is defined in terms of the ground-truth labels it is meant to match, no fitted parameter is renamed as a prediction, and the reference list contains no self-citations that carry the argument. The authors' acknowledged limitation, that the views can only be generated within 'the limited domain in which the views can be generated without increasing the domain shift with the 2D dataset,' identifies an empirical generalization risk about cross-domain transfer from greyscale camera images to LiDAR intensity views; it does not indicate that the results are forced by construction. The lack of an isolated measurement of the 2D model's per-view accuracy on rendered views is a validity concern for the method's premise, but it is not circularity under the definitions used here.

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

The method's central claim rests on empirical transfer assumptions rather than a derived theory. The free parameters are operating points chosen by hand (K, pose noise, depth limits, class merge). No new physical entities are introduced.

free parameters (5)
  • Number of virtual views K = 600
    Ablation shows mIoU drops 86% when K goes from 600 to 100; K=600 is a hand-chosen operating point.
  • Pose randomization ranges theta, lambda, gamma = 30 degrees, 1 m, 1 m
    Set in Section IV-A1; ablation shows high randomization has a strong effect.
  • Intensity clipping range [beta_min, beta_max] = not stated
    Used to normalize intensities to [eta_min, eta_max] in Section III; values not reported, so replication requires guessing.
  • Back-projection depth limits = 1 m to 30 m
    Depth range used for projection and evaluation cropping; affects which points receive votes.
  • Sidewalk/terrain class merge = merged
    Merged 'for this comparison' in Section IV-A1; changes the metric and the pseudo-label task.
assumptions (5)
  • domain assumption A 2D segmentation model trained on greyscale-converted camera images transfers to LiDAR-intensity rendered views.
    Central to the pipeline; relies on augmentations to close the domain gap (Section IV-A1, Fig. 1).
  • domain assumption Only static classes are considered; dynamic objects are excluded because aligned scenes contain motion artifacts.
    Section IV-A1: five classes chosen 'for their static nature'; extension to vehicles and pedestrians left to future work.
  • domain assumption Input sensor poses are available or can be estimated accurately enough for alignment.
    Section III: sensor poses are provided in experiments; KISS-ICP is mentioned but not evaluated.
  • domain assumption 2D annotations for the target classes exist in a public dataset (Mapillary Vistas).
    Training the 2D model requires 2D labels; no 3D labels are used.
  • domain assumption nuScenes and SemanticKITTI share the same set of five static classes of interest.
    The UDA experiment transfers between datasets; class mapping is assumed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of 3D Can Be Explored In 2D: Pseudo-Label Generation for LiDAR Point Clouds Using Sensor-Intensity-Based 2D Semantic Segmentation." pith.science (2026). https://pith.science/paper/SEAXRYDK

@misc{pith2026250503300,
  author       = {Pith},
  title        = {Pith review of: 3D Can Be Explored In 2D: Pseudo-Label Generation for LiDAR Point Clouds Using Sensor-Intensity-Based 2D Semantic Segmentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SEAXRYDK}},
  note         = {Machine review of arXiv:2505.03300}
}
read the original abstract

Semantic segmentation of 3D LiDAR point clouds, essential for autonomous driving and infrastructure management, is best achieved by supervised learning, which demands extensive annotated datasets and faces the problem of domain shifts. We introduce a new 3D semantic segmentation pipeline that leverages aligned scenes and state-of-the-art 2D segmentation methods, avoiding the need for direct 3D annotation or reliance on additional modalities such as camera images at inference time. Our approach generates 2D views from LiDAR scans colored by sensor intensity and applies 2D semantic segmentation to these views using a camera-domain pretrained model. The segmented 2D outputs are then back-projected onto the 3D points, with a simple voting-based estimator that merges the labels associated to each 3D point. Our main contribution is a global pipeline for 3D semantic segmentation requiring no prior 3D annotation and not other modality for inference, which can be used for pseudo-label generation. We conduct a thorough ablation study and demonstrate the potential of the generated pseudo-labels for the Unsupervised Domain Adaptation task.

Figures

Figures reproduced from arXiv: 2505.03300 by the authors.

Figure 1
Figure 1. Pipeline. As a preliminary step, a 2D semantic segmentation model is trained on augmented camera images. An input sequence of LiDAR 3D [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Comparison of the pseudo-labels generated on the nuScenes dataset [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 21 canonical work pages

  1. [6]

    T-UDA: Temporal Unsupervised Domain Adaptation in Sequential Point Clouds

    Awet Haileslassie Gebrehiwot, David Hurych, Karel Zimmermann, Patrick P ´erez, and Tom ´aˇs Svoboda. T-UDA: Temporal Unsupervised Domain Adaptation in Sequential Point Clouds. In 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages 7643–7650, October 2023. ISSN: 2153-0866

    work page 2023

  2. [1]

    SemanticKITTI: A Dataset for Semantic Scene Understanding of LiDAR Sequences, August 2019

    Jens Behley, Martin Garbade, Andres Milioto, Jan Quenzel, Sven Behnke, Cyrill Stachniss, and Juergen Gall. SemanticKITTI: A Dataset for Semantic Scene Understanding of LiDAR Sequences, August 2019. arXiv:1904.01416 [cs]

    arXiv 2019

  3. [2]

    Lang, Sourabh V ora, Venice Erin Liong, Qiang Xu, Anush Krishnan, Yu Pan, Giancarlo Baldan, and Oscar Beijbom

    Holger Caesar, Varun Bankiti, Alex H. Lang, Sourabh V ora, Venice Erin Liong, Qiang Xu, Anush Krishnan, Yu Pan, Giancarlo Baldan, and Oscar Beijbom. nuScenes: A multimodal dataset for autonomous driving, May 2020. arXiv:1903.11027 [cs, stat]

    arXiv 2020

  4. [3]

    Schwing, Alexander Kir- illov, and Rohit Girdhar

    Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kir- illov, and Rohit Girdhar. Masked-attention Mask Transformer for Universal Image Segmentation, June 2022. arXiv:2112.01527 [cs]

    arXiv 2022

  5. [4]

    4D Spatio- Temporal ConvNets: Minkowski Convolutional Neural Networks, June

    Christopher Choy, JunYoung Gwak, and Silvio Savarese. 4D Spatio- Temporal ConvNets: Minkowski Convolutional Neural Networks, June

    work page

  6. [5]

    Corral-Soto, Mrigank Rochan, Yannis Y

    Eduardo R. Corral-Soto, Mrigank Rochan, Yannis Y . He, Xingxin Chen, Shubhra Aich, and Liu Bingbing. Domain Adaptation in LiDAR Semantic Segmentation via Hybrid Learning with Alternating Skip Connections. In 2023 IEEE Intelligent Vehicles Symposium (IV), pages 1–7, Anchorage, AK, USA, June 2023. IEEE

    work page 2023

  7. [7]

    Learning 3D Semantic Segmentation with only 2D Image Supervision

    Kyle Genova, Xiaoqi Yin, Abhijit Kundu, Caroline Pantofaru, For- rester Cole, Avneesh Sud, Brian Brewington, Brian Shucker, and Thomas Funkhouser. Learning 3D Semantic Segmentation with only 2D Image Supervision, October 2021. arXiv:2110.11325 [cs]

    work page Pith review arXiv 2021

  8. [8]

    Seg- ment3D: Learning Fine-Grained Class-Agnostic 3D Segmentation without Manual Labels

    Rui Huang, Songyou Peng, Ayca Takmaz, Federico Tombari, Marc Pollefeys, Shiji Song, Gao Huang, and Francis Engelmann. Seg- ment3D: Learning Fine-Grained Class-Agnostic 3D Segmentation without Manual Labels. 2023. Publisher: arXiv Version Number: 1

    work page 2023

Show all 35 references
  1. [9]

    Berg, Wan-Yen Lo, Piotr Doll ´ar, and Ross Girshick

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexan- der C. Berg, Wan-Yen Lo, Piotr Doll ´ar, and Ross Girshick. Segment Anything, April 2023. arXiv:2304.02643 [cs]

    2023 arXiv

  2. [10]

    Sliced Wasserstein Discrepancy for Unsupervised Domain Adaptation, March 2019

    Chen-Yu Lee, Tanmay Batra, Mohammad Haris Baig, and Daniel Ulbricht. Sliced Wasserstein Discrepancy for Unsupervised Domain Adaptation, March 2019

    2019

  3. [11]

    Pseudo-Label : The Simple and Efficient Semi- Supervised Learning Method for Deep Neural Networks

    Dong-Hyun Lee. Pseudo-Label : The Simple and Efficient Semi- Supervised Learning Method for Deep Neural Networks. ICML 2013 Workshop : Challenges in Representation Learning (WREPL) , July 2013

    2013

  4. [12]

    Segment Any Point Cloud Sequences by Distilling Vision Foundation Models, October 2023

    Youquan Liu, Lingdong Kong, Jun Cen, Runnan Chen, Wenwei Zhang, Liang Pan, Kai Chen, and Ziwei Liu. Segment Any Point Cloud Sequences by Distilling Vision Foundation Models, October 2023. arXiv:2306.09347 [cs]

    2023 arXiv

  5. [13]

    Swin Transformer: Hierarchical Vision Transformer using Shifted Windows, August 2021

    Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, and Baining Guo. Swin Transformer: Hierarchical Vision Transformer using Shifted Windows, August 2021. arXiv:2103.14030 [cs]

    2021 arXiv

  6. [14]

    See More and Know More: Zero-shot Point Cloud Segmentation via Multi-modal Visual Data

    Yuhang Lu, Qi Jiang, Runnan Chen, Yuenan Hou, Xinge Zhu, and Yuexin Ma. See More and Know More: Zero-shot Point Cloud Segmentation via Multi-modal Visual Data. In 2023 IEEE/CVF International Conference on Computer Vision (ICCV) , pages 21617– 21627, Paris, France, October 2023. IEEE

    2023

  7. [15]

    Diffuser: Multi- View 2D-to-3D Label Diffusion for Semantic Scene Segmentation

    Ruben Mascaro, Lucas Teixeira, and Margarita Chli. Diffuser: Multi- View 2D-to-3D Label Diffusion for Semantic Scene Segmentation. In 2021 IEEE International Conference on Robotics and Automation (ICRA), pages 13589–13595, Xi’an, China, May 2021. IEEE

    2021

  8. [16]

    SALUDA: Surface-based Au- tomotive Lidar Unsupervised Domain Adaptation, November 2023

    Bjoern Michele, Alexandre Boulch, Gilles Puy, Tuan-Hung Vu, Re- naud Marlet, and Nicolas Courty. SALUDA: Surface-based Au- tomotive Lidar Unsupervised Domain Adaptation, November 2023. arXiv:2304.03251 [cs]

    2023 arXiv

  9. [17]

    OpenMMLab’s Next-generation Plat- form for General 3D Object Detection, July 2020

    MMDetection3D Contributors. OpenMMLab’s Next-generation Plat- form for General 3D Object Detection, July 2020. original-date: 2020- 07-08T03:39:45Z

    2020

  10. [18]

    OpenMMLab Semantic Segmenta- tion Toolbox and Benchmark, July 2020

    MMSegmentation Contributors. OpenMMLab Semantic Segmenta- tion Toolbox and Benchmark, July 2020. original-date: 2020-06- 14T04:32:33Z

    2020

  11. [19]

    The Mapillary Vistas Dataset for Semantic Understand- ing of Street Scenes

    Gerhard Neuhold, Tobias Ollmann, Samuel Rota Bulo, and Peter Kontschieder. The Mapillary Vistas Dataset for Semantic Understand- ing of Street Scenes. In 2017 IEEE International Conference on Computer Vision (ICCV) , pages 5000–5009, Venice, October 2017. IEEE

    2017

  12. [20]

    Imaging today, foreseeing tomorrow

    Eug ´enio Pellis, Arnadi Murtiyoso, Andrea Masiero, Grazia Tucci, Michele Betti, and Pierre Grussenmeyer. 2D to 3D Label propagation for the semantic segmentation of Heritage building point clouds. In XXIV ISPRS Congress “Imaging today, foreseeing tomorrow”, Commission II 2022...

    2022

  13. [21]

    Semantic segmentation of mobile mapping point clouds via multi-view label transfer

    Torben Peters, Claus Brenner, and Konrad Schindler. Semantic segmentation of mobile mapping point clouds via multi-view label transfer. ISPRS Journal of Photogrammetry and Remote Sensing , 202:30–39, August 2023

    2023

  14. [22]

    Learning Multi- View Aggregation In the Wild for Large-Scale 3D Semantic Segmen- tation, July 2022

    Damien Robert, Bruno Vallet, and Loic Landrieu. Learning Multi- View Aggregation In the Wild for Large-Scale 3D Semantic Segmen- tation, July 2022. arXiv:2204.07548 [cs]

    2022 arXiv

  15. [23]

    Corral-Soto, Amir Nabatchian, and Bingbing Liu

    Mrigank Rochan, Shubhra Aich, Eduardo R. Corral-Soto, Amir Nabatchian, and Bingbing Liu. Unsupervised Domain Adaptation in LiDAR Semantic Segmentation with Self-Supervision and Gated Adapters, March 2022. arXiv:2107.09783 [cs] version: 3

    2022 arXiv

  16. [24]

    Image-to-Lidar Self- Supervised Distillation for Autonomous Driving Data, March 2022

    Corentin Sautier, Gilles Puy, Spyros Gidaris, Alexandre Boulch, Andrei Bursuc, and Renaud Marlet. Image-to-Lidar Self- Supervised Distillation for Autonomous Driving Data, March 2022. arXiv:2203.16258 [cs]

    2022 arXiv

  17. [25]

    Multi-view classification with convolutional neural networks

    Marco Seeland and Patrick M ¨ader. Multi-view classification with convolutional neural networks. PLOS ONE, 16(1):e0245230, January

  18. [26]

    Deep CORAL: Correlation Alignment for Deep Domain Adaptation, July 2016

    Baochen Sun and Kate Saenko. Deep CORAL: Correlation Alignment for Deep Domain Adaptation, July 2016. arXiv:1607.01719 [cs]

    2016 arXiv

  19. [27]

    KISS-ICP: In Defense of Point-to-Point ICP – Simple, Accurate, and Robust Registration If Done the Right Way

    Ignacio Vizzo, Tiziano Guadagnino, Benedikt Mersch, Louis Wies- mann, Jens Behley, and Cyrill Stachniss. KISS-ICP: In Defense of Point-to-Point ICP – Simple, Accurate, and Robust Registration If Done the Right Way. IEEE Robotics and Automation Letters , 8(2):1029–1036, Februar...

    2023 arXiv

  20. [28]

    ADVENT: Adversarial Entropy Minimization for Domain Adaptation in Semantic Segmentation, April 2019

    Tuan-Hung Vu, Himalaya Jain, Maxime Bucher, Matthieu Cord, and Patrick P ´erez. ADVENT: Adversarial Entropy Minimization for Domain Adaptation in Semantic Segmentation, April 2019. arXiv:1811.12833 [cs]

    2019 arXiv

  21. [29]

    Wang, Wei-Lun Chao, Yan Wang, Bharath Hariharan, Kil- ian Q

    Brian H. Wang, Wei-Lun Chao, Yan Wang, Bharath Hariharan, Kil- ian Q. Weinberger, and Mark Campbell. LDLS: 3-D Object Segmen- tation Through Label Diffusion From 2-D Images. IEEE Robotics and Automation Letters, 4(3):2902–2909, July 2019. arXiv:1910.13955 [cs, eess]

    2019 arXiv

  22. [30]

    Transferring CLIP’s Knowledge into Zero-Shot Point Cloud Semantic Segmentation, December 2023

    Yuanbin Wang, Shaofei Huang, Yulu Gao, Zhen Wang, Rui Wang, Kehua Sheng, Bo Zhang, and Si Liu. Transferring CLIP’s Knowledge into Zero-Shot Point Cloud Semantic Segmentation, December 2023. arXiv:2312.07221 [cs]

    2023 arXiv

  23. [31]

    Point Transformer V3: Simpler, Faster, Stronger, December 2023

    Xiaoyang Wu, Li Jiang, Peng-Shuai Wang, Zhijian Liu, Xihui Liu, Yu Qiao, Wanli Ouyang, Tong He, and Hengshuang Zhao. Point Transformer V3: Simpler, Faster, Stronger, December 2023. arXiv:2312.10035 [cs] version: 1

    2023 arXiv

  24. [32]

    ST3D: Self-training for Unsupervised Domain Adaptation on 3D Object Detection, March 2021

    Jihan Yang, Shaoshuai Shi, Zhe Wang, Hongsheng Li, and Xiaojuan Qi. ST3D: Self-training for Unsupervised Domain Adaptation on 3D Object Detection, March 2021. arXiv:2103.05346 [cs]

    2021 arXiv

  25. [33]

    SAM3D: Segment Anything in 3D Scenes, June 2023

    Yunhan Yang, Xiaoyang Wu, Tong He, Hengshuang Zhao, and Xi- hui Liu. SAM3D: Segment Anything in 3D Scenes, June 2023. arXiv:2306.03908 [cs]

    2023 arXiv

  26. [2019]

    arXiv:1904.08755 [cs] version: 4

    1904 arXiv

  27. [2021]

    Publisher: Public Library of Science

Pith tools

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