pith. sign in

arxiv: 2606.20189 · v3 · pith:P26XYSKUnew · submitted 2026-06-18 · 💻 cs.CV · cs.AI· cs.RO

HilDA: Hierarchical Distillation with Diffusion for Advancing Self-Supervised LiDAR Pre-training

Maciej Wozniak , Jesper Ericsson , Hariprasath Govindarajan , Truls Nyberg , Thomas Gustafsson , Patric Jensfelt , Olov Andersson This is my paper

Pith reviewed 2026-06-26 18:04 UTC · model grok-4.3

classification 💻 cs.CV cs.AIcs.RO
keywords LiDAR pre-trainingself-supervised learningknowledge distillationdiffusion modelsautonomous driving3D object detectionscene flowsemantic occupancy
0
0 comments X

The pith

HilDA pre-trains LiDAR backbones by distilling multi-layer semantics and global context from vision models plus a temporal occupancy diffusion task.

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

The paper sets out to demonstrate that current camera-to-LiDAR distillation methods waste information by treating vision foundation models as black boxes and ignoring layer-wise structure, scene-level semantics, and the temporal nature of LiDAR sequences. HilDA instead applies progressive multi-layer distillation, adds global context distillation, and trains with a diffusion objective on temporal occupancy to enforce spatiotemporal consistency. If the approach works, LiDAR models could acquire richer representations for driving scenes from unlabeled data alone. This matters because annotated 3D data remains scarce relative to the geometric variety encountered in real autonomous driving. The central objects carrying the argument are the hierarchical distillation pathway and the diffusion-based occupancy objective.

Core claim

HilDA is a self-supervised pretraining framework for LiDAR backbones that combines hierarchical distillation—multi-layer distillation for progressive semantic alignment and global context distillation for scene-level semantics—with a temporal occupancy diffusion objective to promote spatiotemporal consistency, thereby capturing the semantic what and geometric where needed for downstream driving tasks.

What carries the argument

Hierarchical distillation (multi-layer plus global context) paired with a temporal occupancy diffusion objective.

If this is right

  • Models pre-trained with HilDA reach state-of-the-art numbers on cross-modal distillation benchmarks.
  • They outperform earlier distillation methods on 3D object detection.
  • They improve results on scene flow estimation.
  • They raise performance on semantic occupancy prediction.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same hierarchical-plus-diffusion pattern could be tested on other unpaired sensor pairs such as radar-to-camera.
  • Adding an explicit rigid-body transform step inside the distillation pipeline might further reduce the modality gap the current method leaves implicit.
  • The diffusion objective on occupancy could be replaced by other generative consistency losses to isolate which component drives the reported gains.

Load-bearing premise

Semantic structures and global context extracted from vision foundation models transfer usefully to raw LiDAR geometry without explicit coordinate-frame alignment or special treatment of modality-specific noise.

What would settle it

A controlled experiment in which LiDAR backbones pre-trained with HilDA show no accuracy gain over prior distillation baselines on standard 3D object detection or scene-flow benchmarks would falsify the central claim.

Figures

Figures reproduced from arXiv: 2606.20189 by Hariprasath Govindarajan, Jesper Ericsson, Maciej Wozniak, Olov Andersson, Patric Jensfelt, Thomas Gustafsson, Truls Nyberg.

Figure 1
Figure 1. Figure 1: Effect of HilDA components on semantic accuracy. From baseline (left), we can see error reduction by adding (a) temporal occupancy diffusion, (b) multi-layer distillation, and (c) global context distillation. Best viewed zoomed in Û. Distillation (KD) from Vision Foundation Models (VFMs) into self-supervised LiDAR backbones as a pre-training step [24, 49, 63, 88, 90]. By using VFMs as a teacher, recent wor… view at source ↗
Figure 2
Figure 2. Figure 2: Overview of HilDA. With LiDAR sweeps and synchronized multi-view images, HilDA learns a 3D backbone using three self-supervised objectives: (1) multi-layer point–pixel distillation from a frozen VFM, (2) global context distillation via CLS tokens (t−1 and t0 are processed independently through the distillation modules), and (3) temporal occupancy diffusion conditioned on the student features (F t−1 L , F t… view at source ↗
Figure 3
Figure 3. Figure 3: Qualitative comparison of segmentation performance of HilDA and ScaLR. HilDA makes fewer errors and correctly segments rare cases like a scooter driver (Scene 1) or a person on top of a truck (Scene 2) [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative comparison of 3DOD. Green boxes are GT, and red are predictions. Semantic Occupancy. We evaluate 3D/4D semantic occupancy following Occ4cast [48], with added class labels. The pre-trained backbone is frozen and only a lightweight decoder is trained. Inputs are t−1 and t0 (∆t = 0.5s). We predict occupancy at t0 for 3D, and at t0 and t1 for 4D, and report frame-averaged metrics [PITH_FULL_IMAGE:… view at source ↗
Figure 5
Figure 5. Figure 5: Semantic mIoU over forecast hori￾zon. Decoder trained from t0 through t10 [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Qualitative scene flow com￾parison. Scene Flow. We integrate MinkUnet34 into SSF [37] and evaluate scene flow on Argoverse 2 [82], jointly testing task transfer and domain generalization. Replacing the randomly initialized backbone with HilDA pre-trained weights improves all metrics in Tab. 6, especially dynamic object estimation [PITH_FULL_IMAGE:figures/full_fig_p012_6.png] view at source ↗
read the original abstract

Leveraging Vision Foundation Models (VFMs) for camera-to-LiDAR knowledge distillation offers a promising solution to the scarcity of annotated data needed to represent the immense geometric and kinematic diversity of real-world autonomous driving (AD). However, current approaches typically treat VFMs as black-box teachers, relying exclusively on frame-wise feature similarity. Consequently, they do not fully exploit the teacher's layer-wise semantic structure and global context, as well as the rich spatiotemporal information inherent in LiDAR sequences. We propose HilDA, a self-supervised pretraining framework for LiDAR backbones that better captures the semantic what and geometric where needed for driving tasks. HilDA combines hierarchical distillation comprising multi-layer distillation for progressive semantic alignment and global context distillation for scene-level semantics, with a temporal occupancy diffusion objective promoting spatiotemporal consistency. Models pre-trained with HilDA achieve state-of-the-art results on cross-modal distillation benchmarks and outperform models trained via prior distillation approaches on 3D object detection, scene flow, and semantic occupancy prediction. Code available at: https://maxiuw.github.io/hilda.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper proposes HilDA, a self-supervised pre-training framework for LiDAR backbones that performs hierarchical distillation from vision foundation models (VFMs). It combines multi-layer distillation for progressive semantic alignment, global context distillation for scene-level semantics, and a temporal occupancy diffusion objective for spatiotemporal consistency. The central claim is that models pre-trained with HilDA achieve state-of-the-art results on cross-modal distillation benchmarks and outperform prior distillation methods on downstream tasks including 3D object detection, scene flow, and semantic occupancy prediction.

Significance. If the empirical claims hold after addressing the noted concerns, the work would meaningfully advance self-supervised LiDAR pre-training by more fully exploiting layer-wise structure and global context from VFMs rather than treating them as black boxes. The public code release is a clear strength that supports reproducibility and follow-up work.

major comments (2)
  1. [Abstract, §3] Abstract and §3 (Hierarchical Distillation): The transfer of layer-wise semantic structure and global context is described as relying on feature similarity alone, with no explicit treatment of camera-LiDAR coordinate frame calibration, projection geometry, or modality-specific noise (sparsity, range-dependent density). This assumption is load-bearing for the cross-modal distillation claim; without it, performance gains may not generalize.
  2. [§4, Tables 2-3] §4 (Experiments) and Table 2/3: The SOTA and downstream-task superiority claims require verification that gains are not driven by unablated factors such as training schedule differences or post-hoc hyperparameter choices; the abstract-only review leaves open whether ablations isolate the contribution of the hierarchical components versus the diffusion objective.
minor comments (2)
  1. [§3.3] Notation for the temporal occupancy diffusion loss (Eq. 7 or equivalent) should explicitly define the conditioning variables and the diffusion schedule to avoid ambiguity with standard DDPM formulations.
  2. [Figure 2] Figure 2 (overview diagram) would benefit from clearer indication of how multi-layer features are aligned across modalities before the distillation loss is applied.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. The comments highlight important aspects of cross-modal transfer and experimental rigor that we address point-by-point below. Where revisions are needed to strengthen clarity or provide additional verification, we indicate them explicitly.

read point-by-point responses

  1. Referee: [Abstract, §3] Abstract and §3 (Hierarchical Distillation): The transfer of layer-wise semantic structure and global context is described as relying on feature similarity alone, with no explicit treatment of camera-LiDAR coordinate frame calibration, projection geometry, or modality-specific noise (sparsity, range-dependent density). This assumption is load-bearing for the cross-modal distillation claim; without it, performance gains may not generalize.

    Authors: We appreciate this observation. The full manuscript in §3 describes the use of provided extrinsic calibration matrices to establish point-to-pixel correspondences for feature extraction from VFMs, with LiDAR points projected into image space prior to computing similarity losses. Modality noise is partially mitigated via the temporal occupancy diffusion term, which enforces consistency across frames with varying density. However, we agree that an explicit discussion of projection geometry, handling of out-of-view points, and range-dependent sparsity effects is currently insufficient. We will revise §3 to add a dedicated subsection on these implementation details, including pseudocode for the projection step and a brief analysis of sensitivity to calibration error. revision: yes

  2. Referee: [§4, Tables 2-3] §4 (Experiments) and Table 2/3: The SOTA and downstream-task superiority claims require verification that gains are not driven by unablated factors such as training schedule differences or post-hoc hyperparameter choices; the abstract-only review leaves open whether ablations isolate the contribution of the hierarchical components versus the diffusion objective.

    Authors: The full paper already reports controlled ablations in Tables 2 and 3 that hold training schedule, optimizer, and data fixed while varying only the distillation components (multi-layer, global context) and the diffusion objective. These tables demonstrate incremental gains from each element. To further address potential concerns about hyperparameter sensitivity, we will add a supplementary table in the revision that re-runs all compared methods under identical hyperparameter settings and reports the resulting performance deltas. This will provide stronger isolation of the proposed contributions. revision: partial

Circularity Check

0 steps flagged

No circularity in derivation; empirical claims on external benchmarks

full rationale

The paper describes a self-supervised pre-training method (multi-layer distillation + global context + temporal occupancy diffusion) evaluated on cross-modal benchmarks and downstream tasks (3D detection, scene flow, occupancy). No equations, fitted parameters renamed as predictions, or self-citation chains are present in the provided text that would reduce the claimed SOTA results to inputs by construction. The framework is a standard distillation setup with external validation, so the derivation chain is self-contained.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only; no explicit free parameters, axioms, or invented entities are described. The framework implicitly assumes that vision foundation model features are semantically aligned with LiDAR geometry.

pith-pipeline@v0.9.1-grok · 5745 in / 1124 out tokens · 18545 ms · 2026-06-26T18:04:14.805822+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

107 extracted references · 16 canonical work pages

  1. [1]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2024)

    Agro, B., Sykora, Q., Casas, S., Gilles, T., Urtasun, R.: UnO: Unsupervised occupancy fields for perception and forecasting. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2024)

    2024

  2. [2]

    IEEE Open Journal of Signal Processing (2026).https: //doi.org/10.1109/OJSP.2025.3650437

    Ahn, S., Lee, D., Park, J.: Adaptability of vision foundation models for 3d medical image segmentation. IEEE Open Journal of Signal Processing (2026).https: //doi.org/10.1109/OJSP.2025.3650437

    work page doi:10.1109/ojsp.2025.3650437 2026

  3. [3]

    In: NeurIPS

    Austin, J., Johnson, D.D., Ho, J., Tarlow, D., van den Berg, R.: Structured denoising diffusion models in discrete state-spaces. In: NeurIPS. vol. 34, pp. 17981– 17993 (2021)

    2021

  4. [4]

    In: Proceedings of the International Conference on Computer Vision (2019)

    Behley, J., Garbade, M., Milioto, A., Quenzel, J., Behnke, S., Stachniss, C., Gall, J.: SemanticKITTI: A dataset for semantic scene understanding of LiDAR sequences. In: Proceedings of the International Conference on Computer Vision (2019)

    2019

  5. [5]

    In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2023)

    Blattmann, A., Rombach, R., Ling, H., Dockhorn, T., Kim, S.W., Fidler, S., Kreis, K.: Align your latents: High-resolution video synthesis with latent diffusion models. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2023)

    2023

  6. [6]

    In: Advances in Neural Information Processing Systems (2025)

    Bolya, D., Huang, P.Y., Sun, P., Cho, J.H., Madotto, A., Wei, C., Ma, T., Zhi, J., Rajasegaran, J., Bangalath, H., Wang, J., Monteiro, M., Xu, H., Dong, S., Ravi, N., Li, S.W., Dollár, P., Feichtenhofer, C.: Perception encoder: The best visual embeddings are not at the output of the network. In: Advances in Neural Information Processing Systems (2025)

    2025

  7. [7]

    Boulch, A., Sautier, C., Michele, B., Puy, G., Marlet, R.: ALSO: Automotive LiDARself-supervisionbyoccupancyestimation.In:ProceedingsoftheIEEE/CVF Conference on Computer Vision and Pattern Recognition (2023)

    2023

  8. [8]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2020)

    Caesar, H., Bankiti, V., Lang, A.H., Vora, S., Liong, V.E., Xu, Q., Krishnan, A., Pan, Y., Baldan, G., Beijbom, O.: nuScenes: A multimodal dataset for autonomous driving. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2020)

    2020

  9. [9]

    In: Proceedings of the International Conference on Computer Vision

    Caron, M., Touvron, H., Misra, I., Jégou, H., Mairal, J., Bojanowski, P., Joulin, A.: Emerging properties in self-supervised vision transformers. In: Proceedings of the International Conference on Computer Vision. pp. 9650–9660 (2021)

    2021

  10. [10]

    Data-Centric Engineering p

    Chan, P.H., Li, B., Baris, G., Sadiq, Q., Donzella, V.: The inconvenient truth of ground truth errors in automotive datasets and DNN-based detection. Data-Centric Engineering p. e34 (2024) 16 M. Wozniak et al. HilDA: Hierarchical Distillation with Diffusion

    2024

  11. [11]

    End-to-End Autonomous Driving: Challenges and Frontiers

    Chen, L., Wu, P., Chitta, K., Jaeger, B., Geiger, A., Li, H.: End-to-end autonomous driving: Challenges and frontiers. IEEE Transactions on Pattern Analysis and Machine Intelligence (2024).https://doi.org/10.1109/TPAMI.2024.3435937

    work page doi:10.1109/tpami.2024.3435937 2024

  12. [12]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Chen, P., Liu, S., Zhao, H., Jia, J.: Distilling knowledge via knowledge review. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 5008–5017 (2021)

    2021

  13. [13]

    In: Proceedings of the International Conference on Learning Representations (ICLR) (2025)

    Chen, X., Liu, Z., Xie, S., He, K.: Deconstructing denoising diffusion models for self-supervised learning. In: Proceedings of the International Conference on Learning Representations (ICLR) (2025)

    2025

  14. [14]

    In: Pro- ceedings of the IEEE Winter Conference on Applications of Computer Vision

    Chodosh, N., Ramanan, D., Lucey, S.: Re-evaluating lidar scene flow. In: Pro- ceedings of the IEEE Winter Conference on Applications of Computer Vision. pp. 6005–6015 (January 2024)

    2024

  15. [15]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2019)

    Choy, C., Gwak, J., Savarese, S.: 4D spatio-temporal ConvNets: Minkowski con- volutional neural networks. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2019)

    2019

  16. [16]

    Contributors,P.:Pointcept:Acodebaseforpointcloudperceptionresearch.GitHub repository (2023)

    2023

  17. [17]

    Proceedings of the International Conference on Learning Representations (2024)

    Darcet, T., Oquab, M., Mairal, J., Bojanowski, P.: Vision transformers need registers. Proceedings of the International Conference on Learning Representations (2024)

    2024

  18. [18]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Diehl, C., Sykora, Q., Agro, B., Gilles, T., Casas, S., Urtasun, R.: Dio: Decompos- able implicit 4d occupancy-flow world model. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 27456–27466 (2025)

    2025

  19. [19]

    In: Proceedings of the International Conference on Learning Representations (2021)

    Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al.: An image is worth 16x16 words: Transformers for image recognition at scale. In: Proceedings of the International Conference on Learning Representations (2021)

    2021

  20. [20]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    El Banani, M., Raj, A., Maninis, K.K., Kar, A., Li, Y., Rubinstein, M., Sun, D., Guibas, L., Johnson, J., Jampani, V.: Probing the 3d awareness of visual foundation models. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 21795–21806 (2024)

    2024

  21. [21]

    IEEE Robotics and Automation Letters5(2), 1931–1938 (2020)

    Fang, J., Zhou, D., Yan, F., Zhao, T., Zhang, F., Ma, Y., Wang, L., Yang, R.: Aug- mented lidar simulator for autonomous driving. IEEE Robotics and Automation Letters5(2), 1931–1938 (2020)

    1931

  22. [22]

    In: Proceedings of the International Conference on Computer Vision

    Fu, H., Zhang, D., Zhao, Z., Cui, J., Liang, D., Zhang, C., Zhang, D., Xie, H., Wang, B., Bai, X.: Orion: A holistic end-to-end autonomous driving framework by vision-language instructed action generation. In: Proceedings of the International Conference on Computer Vision. pp. 24823–24834. IEEE (2025)

    2025

  23. [23]

    Vision meets robotics: The kitti dataset.Int

    Geiger, A., Lenz, P., Stiller, C., Urtasun, R.: Vision meets robotics: The KITTI dataset. The International Journal of Robotics Research (2013).https://doi. org/10.1177/0278364913491297

    work page doi:10.1177/0278364913491297 2013

  24. [24]

    In: 36th British Machine Vision Conference 2025, BMVC 2025, Sheffield, UK, November 24-27, 2025

    Govindarajan, H., Wozniak, M., Klingner, M., Maurice, C., Kiran, B.R., Yogamani, S.: Cleverdistiller: Simple and spatially consistent cross-modal distillation. In: 36th British Machine Vision Conference 2025, BMVC 2025, Sheffield, UK, November 24-27, 2025. BMVA (2025)

    2025

  25. [25]

    In: NeurIPS (2020)

    Grill, J.B., Strub, F., Altché, F., Tallec, C., Richemond, P., Buchatskaya, E., Doersch, C., Avila Pires, B., Guo, Z., Gheshlaghi Azar, M., et al.: Bootstrap your own latent-a new approach to self-supervised learning. In: NeurIPS (2020)

    2020

  26. [26]

    arXiv preprint arXiv:2410.10429 (2024) HilDA: Hierarchical Distillation with Diffusion for LiDAR Pre-training 17

    Gu, S., Yin, W., Jin, B., Guo, X., Wang, J., Li, H., Zhang, Q., Long, X.: Dome: Taming diffusion model into high-fidelity controllable occupancy world model. arXiv preprint arXiv:2410.10429 (2024) HilDA: Hierarchical Distillation with Diffusion for LiDAR Pre-training 17

    arXiv 2024

  27. [27]

    In: The Thirteenth International Conference on Learning Representations (2025)

    He, J., Li, H., Yin, W., Liang, Y., Li, L., Zhou, K., Zhang, H., Liu, B., Chen, Y.C.: Lotus: Diffusion-based visual foundation model for high-quality dense prediction. In: The Thirteenth International Conference on Learning Representations (2025)

    2025

  28. [28]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2016)

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

    2016

  29. [29]

    5: Improved baselines for agglomerative vision founda- tion models

    Heinrich, G., Ranzinger, M., Yin, H., Lu, Y., Kautz, J., Tao, A., Catanzaro, B., Molchanov, P.: Radiov2. 5: Improved baselines for agglomerative vision founda- tion models. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 22487–22497 (2025)

    2025

  30. [30]

    In: NeurIPS

    Ho, J., Jain, A., Abbeel, P.: Denoising diffusion probabilistic models. In: NeurIPS. vol. 33 (2020)

    2020

  31. [31]

    Hudson, D.A., Zoran, D., Malinowski, M., Lampinen, A.K., Jaegle, A., McClelland, J.L., Matthey, L., Hill, F., Lerchner, A.: Soda: Bottleneck diffusion models for representationlearning.In:ProceedingsoftheIEEE/CVFConferenceonComputer Vision and Pattern Recognition (CVPR). pp. 23115–23127 (2024)

    2024

  32. [32]

    Joska, L

    Jiang, P., Osteen, P., Wigness, M., Saripalli, S.: RELLIS-3D dataset: Data, bench- marks and analysis. In: 2021 IEEE International Conference on Robotics and Au- tomation (ICRA) (2021).https://doi.org/10.1109/ICRA48506.2021.9561251

    work page doi:10.1109/icra48506.2021.9561251 2021

  33. [33]

    In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)

    Justo Miro, A., af Klinteberg, L., Timus, B., Asefaw, A., Khoche, A., Gustafsson, T., Mansouri, S.S., Daneshtalab, M.: Correcting and Quantifying Systematic Errors in 3D Box Annotations for Autonomous Driving. In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV). pp. 6724–6732 (March 2026)

    2026

  34. [34]

    IEEE Robotics and Automation Letters7(2), 1447–1454 (2021) https://doi.org/10.1109/LRA

    Keetha, N., Mishra, A., Karhade, J., Jatavallabhula, K.M., Scherer, S., Krishna, M., Garg, S.: AnyLoc: Towards universal visual place recognition. IEEE Robotics and Automation Letters9(2), 1286–1293 (2024).https://doi.org/10.1109/LRA. 2023.3343602

    work page doi:10.1109/lra 2024

  35. [35]

    Khatri, I., Vedder, K., Peri, N., Ramanan, D., Hays, J.: I Can’t Believe It’s Not Scene Flow! In: Proceedings of the European Conference on Computer Vision (2024).https://doi.org/10.1007/978-3-031-72649-1_14

    work page doi:10.1007/978-3-031-72649-1_14 2024

  36. [36]

    In: 2024 European Control Conference (ECC)

    Khoche, A., Asefaw, A., González, A., Timus, B., Mansouri, S.S., Jensfelt, P.: Addressing data annotation challenges in multiple sensors: A solution for scania collected datasets. In: 2024 European Control Conference (ECC). pp. 1032–1038 (2024)

    2024

  37. [37]

    doi: 10.1109/ICRA55743.2025.11128816

    Khoche, A., Zhang, Q., Sánchez, L.P., Asefaw, A., Mansouri, S.S., Jensfelt, P.: SSF: Sparse long-range scene flow for autonomous driving. In: Proceedings of the IEEE International Conference on Robotics and Automation (2025).https: //doi.org/10.1109/ICRA55743.2025.11128770

    work page doi:10.1109/icra55743.2025.11128770 2025

  38. [38]

    In: Interna- tional Conference on Learning Representations (ICLR) (2015)

    Kingma, D.P., Ba, J.: Adam: A method for stochastic optimization. In: Interna- tional Conference on Learning Representations (ICLR) (2015)

    2015

  39. [39]

    In: Proceedings of the International Conference on Computer Vision (2023)

    Kirillov, A., Mintun, E., Ravi, N., Mao, H., Rolland, C., Gustafson, L., Xiao, T., Whitehead, S., Berg, A.C., Lo, W.Y., et al.: Segment anything. In: Proceedings of the International Conference on Computer Vision (2023)

    2023

  40. [40]

    IEEE Access11, 79341–79356 (2023)

    Klokov, A.A., Pak, D.U., Khorin, A., Yudin, D.A., Kochiev, L., Luchinskiy, V.D., Bezuglyj, V.D.: DAPS3D: Domain adaptive projective segmentation of 3D LiDAR point clouds. IEEE Access11, 79341–79356 (2023)

    2023

  41. [41]

    In: Proceedings of the International Conference on Computer Vision (2023) 18 M

    Kong, L., Liu, Y., Li, X., Chen, R., Zhang, W., Ren, J., Pan, L., Chen, K., Liu, Z.: Robo3D: Towards robust and reliable 3D perception against corruptions. In: Proceedings of the International Conference on Computer Vision (2023) 18 M. Wozniak et al. HilDA: Hierarchical Distillation with Diffusion

    2023

  42. [42]

    GitHub repository (2026), accessed 2026-01-09

    KTH-RPL, contributors: Opensceneflow: A codebase for point cloud scene flow estimation research. GitHub repository (2026), accessed 2026-01-09

    2026

  43. [43]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    Li, B., Guo, J., Liu, H., Zou, Y., Ding, Y., Chen, X., Zhu, H., Tan, F., Zhang, C., Wang, T., Zhou, S., Zhang, L., Qi, X., Zhao, H., Yang, M., Zeng, W., Jin, X.: Uniscene: Unified occupancy-centric driving scene generation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 11971–11981 (June 2025)

    2025

  44. [44]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Li, J., Saltori, C., Poiesi, F., Sebe, N.: Cross-modal and uncertainty-aware ag- glomeration for open-vocabulary 3d scene understanding. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 19390– 19400 (June 2025)

    2025

  45. [45]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Liao, B., Chen, S., Yin, H., Jiang, B., Wang, C., Yan, S., Zhang, X., Li, X., Zhang, Y., Zhang, Q., Wang, X.: Diffusiondrive: Truncated diffusion model for end-to-end autonomous driving. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 12037–12047. IEEE (2025)

    2025

  46. [46]

    arXiv preprint arXiv:1805.06334 (2018)

    Liebel, L., Körner, M.: Auxiliary tasks in multi-task learning. arXiv preprint arXiv:1805.06334 (2018)

    Pith/arXiv arXiv 2018

  47. [47]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Liu, J., Wang, G., Ye, W., Jiang, C., Han, J., Liu, Z., Zhang, G., Du, D., Wang, H.: Difflow3d: Toward robust uncertainty-aware scene flow estimation with iterative diffusion-based refinement. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 15109–15119 (2024)

    2024

  48. [48]

    a fer, Andrew Wing Keung To, Kuan-Ho Lao, Murat Cubuktepe, Matthew Haley, Peter B \

    Liu, X., Gong, M., Fang, Q., Xie, H., Li, Y., Zhao, H., Feng, C.: LiDAR-based 4D occupancy completion and forecasting. In: 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) (2024).https://doi.org/ 10.1109/IROS58592.2024.10801302

    work page doi:10.1109/iros58592.2024.10801302 2024

  49. [49]

    In: Advances in Neural Information Processing Systems

    Liu, Y., Kong, L., Cen, J., Chen, R., Zhang, W., Pan, L., Chen, K., Liu, Z.: Segment any point cloud sequences by distilling vision foundation models. In: Advances in Neural Information Processing Systems. vol. 36 (2023)

    2023

  50. [50]

    arXiv preprint arXiv:2104.04687 (2021)

    Liu, Y.C., Huang, Y.K., Chiang, H.Y., Su, H.T., Liu, Z.Y., Chen, C.T., Tseng, C.Y., Hsu, W.H.: Learning from 2D: Contrastive pixel-to-point knowledge transfer for 3D pretraining. arXiv preprint arXiv:2104.04687 (2021)

    arXiv 2021

  51. [51]

    In: Proceedings of the International Conference on Computer Vision (2025)

    Liu, Y., Fu, J., Wu, Y., Wu, K., Li, P., Wu, J., Zhou, S., Xin, J.: Mind the gap: Aligning vision foundation models to image feature matching. In: Proceedings of the International Conference on Computer Vision (2025)

    2025

  52. [52]

    In: Proceedings of the IEEE Winter Conference on Applications of Computer Vision (2026)

    Ljungbergh, W., Lilja, A., Tonderski, A., Ling, A.L., Lindström, C., Verbeke, W., Fu, J., Petersson, C., Hammarstrand, L., Felsberg, M.: GASP: Unifying geometric and semantic self-supervised pre-training for autonomous driving. In: Proceedings of the IEEE Winter Conference on Applications of Computer Vision (2026)

    2026

  53. [53]

    In: International Conference on Learning Representations (ICLR) (2019)

    Loshchilov, I., Hutter, F.: Decoupled weight decay regularization. In: International Conference on Learning Representations (ICLR) (2019)

    2019

  54. [54]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Ma, J., Chen, X., Huang, J., Xu, J., Luo, Z., Xu, J., Gu, W., Ai, R., Wang, H.: Cam4docc: Benchmark for camera-only 4d occupancy forecasting in autonomous driving applications. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 21486–21495 (2024)

    2024

  55. [55]

    IEEE Transactions on Pattern Analysis and Machine Intelligence44(8), 4454–4468 (2022).https://doi.org/10.1109/TPAMI.2021.3063611

    Meng, Q., Wang, W., Zhou, T., Shen, J., Jia, Y., Van Gool, L.: Towards a weakly supervised framework for 3D point cloud object detection and annotation. IEEE Transactions on Pattern Analysis and Machine Intelligence44(8), 4454–4468 (2022).https://doi.org/10.1109/TPAMI.2021.3063611

    work page doi:10.1109/tpami.2021.3063611 2022

  56. [56]

    MMDetection3D Contributors: MMDetection3D: OpenMMLab next-generation platform for general 3D object detection (2020) HilDA: Hierarchical Distillation with Diffusion for LiDAR Pre-training 19

    2020

  57. [57]

    Murrugarra-Llerena, J., Kirsten, L., Jung, C.R.: Can we trust bounding box annotations for object detection? In: 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). pp. 4812–4821 (2022)

    2022

  58. [58]

    In: International Conference on Machine Learning

    Nichol, A.Q., Dhariwal, P.: Improved denoising diffusion probabilistic models. In: International Conference on Machine Learning. Proceedings of Machine Learning Research, vol. 139, pp. 8162–8171. PMLR (2021)

    2021

  59. [59]

    In: Proceedings of the 35th Conference on Neural Information Processing Systems Track on Datasets and Benchmarks (December 2021)

    Northcutt, C.G., Athalye, A., Mueller, J.: Pervasive label errors in test sets destabilize machine learning benchmarks. In: Proceedings of the 35th Conference on Neural Information Processing Systems Track on Datasets and Benchmarks (December 2021)

    2021

  60. [60]

    In: IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    Nunes, L., Marcuzzi, R., Mersch, B., Behley, J., Stachniss, C.: Scaling diffusion models to real-world 3d lidar scene completion. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 14770–14780. IEEE (2024)

    2024

  61. [61]

    Transactions on Machine Learning Research (2024)

    Oquab, M., Darcet, T., Moutakanni, T., Vo, H.V., Szafraniec, M., Khalidov, V., Fernandez, P., Haziza, D., Massa, F., El-Nouby, A., Assran, M., Ballas, N., Galuba, W., Howes, R., Huang, P.Y., Li, S.W., Misra, I., Rabbat, M., Sharma, V., Synnaeve, G., Xu, H., Jegou, H., Mairal, J., Labatut, P., Joulin, A., Bojanowski, P.: DINOv2: Learning robust visual feat...

    2024

  62. [62]

    Otaduy, and Dan Casas

    Preechakul, K., Chatthee, N., Wizadwongsa, S., Suwajanakorn, S.: Diffusion autoencoders: Toward a meaningful and decodable representation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2022).https://doi.org/10.1109/CVPR52688.2022.01036

    work page doi:10.1109/cvpr52688.2022.01036 2022

  63. [63]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2024)

    Puy, G., Gidaris, S., Boulch, A., Siméoni, O., Sautier, C., Pérez, P., Bursuc, A., Marlet, R.: Three pillars improving vision foundation model distillation for LiDAR. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2024)

    2024

  64. [64]

    In: International Conference on Machine Learning (2021)

    Radford, A., Kim, J.W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., Krueger, G., Sutskever, I.: Learning transferable visual models from natural language supervision. In: International Conference on Machine Learning (2021)

    2021

  65. [65]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Ranzinger, M., Heinrich, G., Kautz, J., Molchanov, P.: Am-radio: Agglomerative vision foundation model reduce all domains into one. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 12490– 12500 (2024)

    2024

  66. [66]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Rombach, R., Blattmann, A., Lorenz, D., Esser, P., Ommer, B.: High-resolution image synthesis with latent diffusion models. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 10684–10695 (2022)

    2022

  67. [67]

    In: Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015

    Ronneberger, O., Fischer, P., Brox, T.: U-net: Convolutional networks for biomed- ical image segmentation. In: Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015. Lecture Notes in Computer Science, vol. 9351, pp. 234–241. Springer, Cham (2015)

    2015

  68. [68]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2022)

    Sautier, C., Puy, G., Gidaris, S., Boulch, A., Bursuc, A., Marlet, R.: Image-to- LiDAR self-supervised distillation for autonomous driving data. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2022)

    2022

  69. [69]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2019)

    Shi, S., Wang, X., Li, H.: PointRCNN: 3D object proposal generation and detection from point cloud. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2019)

    2019

  70. [70]

    Wozniak et al

    Sun, P., Kretzschmar, H., Dotiwalla, X., Chouard, A., Patnaik, V., Tsui, P., Guo, J., Zhou, Y., Chai, Y., Caine, B., Vasudevan, V., Han, W., Ngiam, J., Zhao, H., 20 M. Wozniak et al. HilDA: Hierarchical Distillation with Diffusion Timofeev, A., Ettinger, S., Krivokon, M., Gao, A., Joshi, A., Zhang, Y., Shlens, J., Chen, Z., Anguelov, D.: Scalability in pe...

    2020

  71. [71]

    Proceedings of the AAAI Conference on Artificial Intelligence (2026).https://doi.org/10.1609/aaai.v40i11.37913

    Tian, H., Xu, B., Li, S.: Distillation dynamics: Towards understanding feature- based distillation in vision transformers. Proceedings of the AAAI Conference on Artificial Intelligence (2026).https://doi.org/10.1609/aaai.v40i11.37913

    work page doi:10.1609/aaai.v40i11.37913 2026

  72. [72]

    In: International conference on machine learning

    Touvron, H., Cord, M., Douze, M., Massa, F., Sablayrolles, A., Jégou, H.: Training data-efficient image transformers & distillation through attention. In: International conference on machine learning. pp. 10347–10357. PMLR (2021)

    2021

  73. [73]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Unal, O., Dai, D., Van Gool, L.: Scribble-supervised LiDAR semantic segmentation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 2697–2707 (2022)

    2022

  74. [74]

    In: NeurIPS (2022)

    Voleti, V., Jolicoeur-Martineau, A., Pal, C.: Mcvd: Masked conditional video diffusion for prediction, generation, and interpolation. In: NeurIPS (2022)

    2022

  75. [75]

    In: Proceedings of the European Conference on Computer Vision

    Wang, G., Wang, Z., Tang, P., Zheng, J., Ren, X., Feng, B., Ma, C.: Occgen: Generative multi-modal 3d occupancy prediction for autonomous driving. In: Proceedings of the European Conference on Computer Vision. pp. 95–112. Springer (2024)

    2024

  76. [76]

    Drive- dreamer: Towards real-world-drive world models for autonomous driving

    Wang, X., Zhu, Z., Huang, G., Chen, X., Zhu, J., Lu, J.: Drivedreamer: Towards real-world-drive world models for autonomous driving. In: Computer Vision – ECCV 2024 (2025).https://doi.org/10.1007/978-3-031-73195-2_4

    work page doi:10.1007/978-3-031-73195-2_4 2024

  77. [77]

    In: Proceedings of the International Conference on Computer Vision

    Wang, X., Zhu, Z., Xu, W., Zhang, Y., Wei, Y., Chi, X., Ye, Y., Du, D., Lu, J., Wang, X.: Openoccupancy: A large scale benchmark for surrounding semantic oc- cupancy perception. In: Proceedings of the International Conference on Computer Vision. pp. 17850–17859 (October 2023)

    2023

  78. [78]

    arXiv preprint arXiv:2511.00088 (2025)

    Wang, Y., Luo, W., Bai, J., Cao, Y., Che, T., Chen, K., Chen, Y., Diamond, J., Ding, Y., Ding, W., et al.: Alpamayo-r1: Bridging reasoning and action pre- diction for generalizable autonomous driving in the long tail. arXiv preprint arXiv:2511.00088 (2025)

    Pith/arXiv arXiv 2025

  79. [79]

    doi: 10.1109/ICRA55743.2025.11128816

    Wang, Y., Liu, Y., Yuan, T., Mao, Y., Liang, Y., Yang, X., Zhang, H., Zhao, H.: Diffusion-based generative models for 3D occupancy prediction in autonomous driving. In: 2025 IEEE International Conference on Robotics and Automation (ICRA) (2025).https://doi.org/10.1109/ICRA55743.2025.11128716

    work page doi:10.1109/icra55743.2025.11128716 2025

  80. [80]

    In: Proceedings of the International Conference on Computer Vision

    Wei, C., Mangalam, K., Huang, P.Y., Li, Y., Fan, H., Xu, H., Wang, H., Xie, C., Yuille, A.L., Feichtenhofer, C.: Diffusion models as masked autoencoders. In: Proceedings of the International Conference on Computer Vision. pp. 16238–16248 (2023)

    2023

Showing first 80 references.