pith. sign in

arxiv: 2606.21373 · v1 · pith:DTB6VDXOnew · submitted 2026-06-19 · 💻 cs.CV · cs.RO

FLM-Occ: Feed-forward Likelihood Maximization for Efficient Indoor Occupancy Prediction

Guangcheng Chen , Lihuang Fang , Huaqi Tao , Yicheng He , Li He , Hong Zhang This is my paper

Pith reviewed 2026-06-26 14:32 UTC · model grok-4.3

classification 💻 cs.CV cs.RO
keywords occupancy predictionindoor scenesmixture modelssuperquadricslikelihood maximization3D representationvoxel distribution estimationfeed-forward training
0
0 comments X

The pith

FLM-Occ predicts indoor occupancy by maximizing the likelihood of a mixture model over ground-truth occupied voxels.

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

The paper aims to show that reformulating occupancy prediction as a feed-forward likelihood maximization problem allows networks to learn sparse 3D representations with far fewer primitives than before. Previous Gaussian primitive methods suffered from spurious predictions in empty space due to only local voxel classification constraints. By defining mixture weights as normalized primitive volumes and deriving voxelization formulas, FLM enables global supervision that eliminates unnecessary primitives. This leads to more efficient models that can relocate primitives over long distances to fit scenes accurately. A reader would care because it promises both higher accuracy and much lower computational cost for tasks like indoor scene understanding.

Core claim

FLM-Occ is a method based on Feed-forward Likelihood Maximization that trains a network to predict a mixture model maximizing the likelihood over ground-truth occupied voxels. It uses only 32 superquadrics to achieve superior accuracy on Occ-ScanNet compared to prior state-of-the-art, representing just 2.7% of the primitives used before, and runs 3.7 times faster by avoiding spurious primitives in empty regions.

What carries the argument

Feed-forward Likelihood Maximization (FLM), which reformulates occupancy prediction as voxel distribution estimation and uses normalized primitive volumes as mixture weights with derived voxelization formulas to enable end-to-end training.

If this is right

  • Networks can be trained end-to-end to predict mixture models for occupancy without local-only constraints.
  • Primitive count can be reduced dramatically while maintaining or improving accuracy by eliminating empty-space predictions.
  • Primitives are able to relocate over long distances during optimization to model the scene effectively.
  • The approach provides global supervision that previous voxel classification methods lacked.

Where Pith is reading between the lines

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

  • Similar likelihood maximization could improve efficiency in other sparse 3D representation tasks beyond indoor occupancy.
  • Testing FLM with different primitive types like Gaussians or other shapes might yield further gains.
  • Applying the framework to dynamic scenes or real-time applications could leverage the speed improvements.
  • The reduced primitive count suggests potential for scaling to larger environments without proportional compute increase.

Load-bearing premise

That defining mixture weights as normalized primitive volumes and deriving the corresponding voxelization formulas will allow stable end-to-end training while providing sufficient global supervision to eliminate spurious primitives in empty space.

What would settle it

Running FLM-Occ on Occ-ScanNet and checking if it achieves higher accuracy than prior methods with 32 superquadrics or if the number of predicted primitives in empty regions remains high compared to ground truth.

Figures

Figures reproduced from arXiv: 2606.21373 by Guangcheng Chen, Hong Zhang, Huaqi Tao, Li He, Lihuang Fang, Yicheng He.

Figure 1
Figure 1. Figure 1: Comparison (top-left) on Occ-ScanNet [42]. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Four Gaussian-based methods for occupancy prediction. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Information graphs of the training losses for a single primitive via [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: 2D illustrations of the roles of the two terms in the FLM loss. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Overview of our FLM-Occ (left) and the structure of the refinement [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Performance vs. Representation Size. Filled and hollow markers in (a) indicate IoU and mIoU, respectively. FPS is measured on a single RTX 3090 GPU. affect the results on Occ-ScanNet [42]. We also calculate the average distance between the initialized and final primitives. The results are shown in [PITH_FULL_IMAGE:figures/full_fig_p012_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Distribution of density and IoUs with different density thresholds. [PITH_FULL_IMAGE:figures/full_fig_p013_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Comparisons on Occ-ScanNet dataset [42] 4.7 Qualitative Results [PITH_FULL_IMAGE:figures/full_fig_p014_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Fitting shapes with 16 / 256 Gaussians using two losses, respectively. [PITH_FULL_IMAGE:figures/full_fig_p020_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Predicted primitives and voxels by SplatSSC [27] and FLM-Occ. [PITH_FULL_IMAGE:figures/full_fig_p021_10.png] view at source ↗
read the original abstract

Recent indoor occupancy prediction methods adopt Gaussian primitives as a sparse 3D representation for computational efficiency. However, their training relies on voxel classification, which imposes only local constraints and lacks global supervision on the distribution of the primitives. Therefore, they inevitably predict spurious primitives in empty regions, undermining both representational and computational efficiency. To address this, we propose Feed-forward Likelihood Maximization (FLM), a novel framework that reformulates occupancy prediction as voxel distribution estimation. In FLM, a network is trained to predict a mixture model that maximizes the likelihood over ground-truth occupied voxels in a feed-forward manner. To enable end-to-end training of networks and voxelization of a standard mixture model, we define mixture weights as normalized primitive volumes to implicitly enforce simplex constraints and derive novel voxelization formulas. Based on FLM, our FLM-Occ, a novel method that is capable of relocating randomly initialized primitives over long distances to model a scene. On Occ-ScanNet, FLM-Occ achieves superior accuracy using only 32 superquadrics, 2.7% of the prior SoTA, while running 3.7 times faster.

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

1 major / 1 minor

Summary. The manuscript proposes Feed-forward Likelihood Maximization (FLM) to reformulate indoor occupancy prediction as voxel distribution estimation. A network predicts a mixture of superquadrics whose weights are defined as normalized primitive volumes; novel voxelization formulas are derived to enable end-to-end likelihood maximization over ground-truth occupied voxels. The central claim is that this supplies global supervision sufficient to eliminate spurious primitives in empty space. On Occ-ScanNet the resulting FLM-Occ method reportedly achieves superior accuracy with only 32 superquadrics (2.7 % of prior SoTA) while running 3.7 imes faster.

Significance. If the likelihood objective with the stated volume-normalized weights and voxelization formulas does in fact deliver global supervision that suppresses empty-space primitives, the approach would constitute a concrete improvement in the efficiency of sparse 3D representations for indoor scenes.

major comments (1)
  1. [Abstract (FLM definition)] Abstract (FLM definition paragraph): the claim that normalized primitive volumes plus the derived voxelization formulas supply global supervision sufficient to eliminate spurious primitives is load-bearing for the efficiency result. Because superquadrics overlap, the sum of individual volumes exceeds the measure of the occupied set; the manuscript must show that the voxelization formulas incorporate an explicit overlap correction (e.g., inclusion-exclusion, soft union, or proper density integration) so that the likelihood penalizes non-zero mass in empty regions. Absent such a correction the objective remains local and the 32-primitive claim is not supported.
minor comments (1)
  1. [Abstract] Abstract: reported accuracy and runtime figures are given without error bars, statistical significance, or reference to the precise evaluation protocol (e.g., voxel resolution, train/val split).

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the detailed and constructive review. Below we respond point-by-point to the single major comment.

read point-by-point responses

  1. Referee: Abstract (FLM definition paragraph): the claim that normalized primitive volumes plus the derived voxelization formulas supply global supervision sufficient to eliminate spurious primitives is load-bearing for the efficiency result. Because superquadrics overlap, the sum of individual volumes exceeds the measure of the occupied set; the manuscript must show that the voxelization formulas incorporate an explicit overlap correction (e.g., inclusion-exclusion, soft union, or proper density integration) so that the likelihood penalizes non-zero mass in empty regions. Absent such a correction the objective remains local and the 32-primitive claim is not supported.

    Authors: The volume normalization defines mixture weights that sum exactly to one, yielding a valid probability distribution whose integral over all space equals one. The voxelization formulas compute the probability mass of each voxel under this mixture density (i.e., the integral of the weighted sum of primitive densities over the voxel). Because the total probability mass is constrained to one, any mass placed in empty space necessarily reduces the mass available to occupied voxels; the likelihood objective is therefore global by construction. Overlaps are already accounted for by the additive mixture density; no separate inclusion-exclusion term is required. The derivation appears in Section 3.2. We will add a clarifying sentence in the abstract and in Section 3.2 to make this explicit. revision: partial

Circularity Check

0 steps flagged

No significant circularity; derivation uses external GT voxels and independent modeling choices

full rationale

The paper's core step defines mixture weights as normalized primitive volumes to enforce simplex constraints and derives voxelization formulas for end-to-end training. This is an explicit modeling ansatz, not a self-definition or fitted quantity renamed as prediction. The likelihood objective is maximized over external ground-truth occupied voxels, providing an independent training signal. No self-citations appear in the provided text, and no step reduces the claimed accuracy or efficiency result to a tautology by construction. The 32-primitive result is presented as an empirical outcome on Occ-ScanNet, not forced by the weight definition alone.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The approach rests on the domain assumption that a standard mixture model can be voxelized differentiably once weights are tied to primitive volumes, plus standard mathematical properties of mixture models and simplex constraints.

axioms (1)
  • domain assumption Mixture weights defined as normalized primitive volumes implicitly enforce simplex constraints
    Stated in abstract as the device enabling end-to-end training and voxelization.

pith-pipeline@v0.9.1-grok · 5745 in / 1295 out tokens · 25001 ms · 2026-06-26T14:32:19.759520+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

48 extracted references · 3 canonical work pages

  1. [1]

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

    Cao, A.Q., De Charette, R.: Monoscene: Monocular 3d semantic scene completion. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 3991–4001 (2022)

    2022

  2. [2]

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

    Charatan,D.,Li,S.L.,Tagliasacchi,A.,Sitzmann,V.:pixelsplat:3dgaussiansplats from image pairs for scalable generalizable 3d reconstruction. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 19457– 19467 (2024)

    2024

  3. [3]

    In: European Conference on Computer Vision

    Chen, Y., Xu, H., Zheng, C., Zhuang, B., Pollefeys, M., Geiger, A., Cham, T.J., Cai, J.: Mvsplat: Efficient 3d gaussian splatting from sparse multi-view images. In: European Conference on Computer Vision. pp. 370–386. Springer (2024)

    2024

  4. [4]

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

    Dai,A.,Chang,A.X.,Savva,M.,Halber,M.,Funkhouser,T.,Nießner,M.:Scannet: Richly-annotated 3d reconstructions of indoor scenes. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 5828–5839 (2017)

    2017

  5. [5]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Gan, W., Liu, F., Xu, H., Mo, N., Yokoya, N.: Gaussianocc: Fully self-supervised and efficient 3d occupancy estimation with gaussian splatting. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 28980–28990 (2025)

    2025

  6. [6]

    In: 2024 IEEE International Conference on Robotics and Automation (ICRA)

    Hou, J., Li, X., Guan, W., Zhang, G., Feng, D., Du, Y., Xue, X., Pu, J.: Fastocc: Accelerating 3d occupancy prediction by fusing the 2d bird’s-eye view and perspec- tive view. In: 2024 IEEE International Conference on Robotics and Automation (ICRA). pp. 16425–16431. IEEE (2024)

    2024

  7. [7]

    In: Proceedings of the IEEE/CVF International Con- ference on Computer Vision

    Huang, R., Mikolajczyk, K.: No pose at all: Self-supervised pose-free 3d gaussian splatting from sparse views. In: Proceedings of the IEEE/CVF International Con- ference on Computer Vision. pp. 27947–27957 (2025)

    2025

  8. [8]

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

    Huang, Y., Thammatadatrakoon, A., Zheng, W., Zhang, Y., Du, D., Lu, J.: Gaussianformer-2: Probabilistic gaussian superposition for efficient 3d occupancy prediction. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 27477–27486 (June 2025)

    2025

  9. [9]

    2024 , url =

    Huang, Y., Zheng, W., Zhang, B., Zhou, J., Lu, J.: SelfOcc: Self-Supervised Vision-Based 3D Occupancy Prediction . In: 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 19946–19956. IEEE Computer Society, Los Alamitos, CA, USA (Jun 2024).https : / / doi . org / 10.1109/CVPR52733.2024.01885,https://doi.ieeecomputersociety.org/...

    work page doi:10.1109/cvpr52733.2024.01885 2024

  10. [10]

    In: Proceedings of the IEEE/CVF con- ference on computer vision and pattern recognition

    Huang, Y., Zheng, W., Zhang, Y., Zhou, J., Lu, J.: Tri-perspective view for vision- based 3d semantic occupancy prediction. In: Proceedings of the IEEE/CVF con- ference on computer vision and pattern recognition. pp. 9223–9232 (2023) 16 G. Chen, L. Huanget al

    2023

  11. [11]

    In: Leonardis, A., Ricci, E., Roth, S., Russakovsky, O., Sattler, T., Varol, G

    Huang, Y., Zheng, W., Zhang, Y., Zhou, J., Lu, J.: Gaussianformer: Scene as gaus- sians for vision-based 3d semantic occupancy prediction. In: Leonardis, A., Ricci, E., Roth, S., Russakovsky, O., Sattler, T., Varol, G. (eds.) Computer Vision – ECCV 2024. pp. 376–393. Springer Nature Switzerland, Cham (2025)

    2024

  12. [12]

    In: European Conference on Computer Vision

    Iwase, S., Liu, K., Guizilini, V., Gaidon, A., Kitani, K., Ambruş, R., Zakharov, S.: Zero-shot multi-object scene completion. In: European Conference on Computer Vision. pp. 96–113. Springer (2024)

    2024

  13. [13]

    In: IEEE/CVF International Conference on Computer Vision (ICCV) (2025)

    Jevtić, A., Reich, C., Wimbauer, F., Hahn, O., Rupprecht, C., Roth, S., Cre- mers, D.: Feed-forward SceneDINO for unsupervised semantic scene completion. In: IEEE/CVF International Conference on Computer Vision (ICCV) (2025)

    2025

  14. [14]

    ACM Trans

    Kerbl, B., Kopanas, G., Leimkühler, T., Drettakis, G.: 3d gaussian splatting for real-time radiance field rendering. ACM Trans. Graph.42(4), 139–1 (2023)

    2023

  15. [15]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Li, H., Hou, Y., Xing, X., Ma, Y., Sun, X., Zhang, Y.: Occmamba: Semantic occu- pancy prediction with state space models. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 11949–11959 (2025)

    2025

  16. [16]

    In: 2025 IEEE International Conference on Robotics and Automation (ICRA)

    Li, J., Lu, M., Liu, J., Wang, H., Gu, C., Zheng, W., Du, L., Zhang, S.: Sliceocc: Indoor 3d semantic occupancy prediction with vertical slice representation. In: 2025 IEEE International Conference on Robotics and Automation (ICRA). pp. 15762–15768. IEEE (2025)

    2025

  17. [17]

    IEEE Robotics and Automation Letters10(11), 11690–11697 (2025)

    Li, X., Zheng, Y., Li, P., Chen, Y., Zhang, Y.Q., Ding, W.: Enhancing indoor occupancy prediction via sparse query-based multi-level consistent knowledge dis- tillation. IEEE Robotics and Automation Letters10(11), 11690–11697 (2025). https://doi.org/10.1109/LRA.2025.3615532

    work page doi:10.1109/lra.2025.3615532 2025

  18. [18]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2023)

    Li, Y., Yu, Z., Choy, C., Xiao, C., Alvarez, J.M., Fidler, S., Feng, C., Anandkumar, A.: Voxformer: Sparse voxel transformer for camera-based 3d semantic scene com- pletion. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2023)

    2023

  19. [19]

    Lin,X.,Lin,T.,Pei,Z.,Huang,L.,Su,Z.:Sparse4d:Multi-view3dobjectdetection with sparse spatial-temporal fusion (2022)

    2022

  20. [20]

    In: International Conference on Learning Representations (2019)

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

    2019

  21. [21]

    Advances in Neural Information Processing Systems37, 79618–79641 (2024)

    Lu, Y., Zhu, X., Wang, T., Ma, Y.: Octreeocc: Efficient and multi-granularity occu- pancy prediction using octree queries. Advances in Neural Information Processing Systems37, 79618–79641 (2024)

    2024

  22. [22]

    In: ProceedingsoftheIEEE/CVFConferenceonComputerVisionandPatternRecog- nition

    Matsuki, H., Murai, R., Kelly, P.H., Davison, A.J.: Gaussian splatting slam. In: ProceedingsoftheIEEE/CVFConferenceonComputerVisionandPatternRecog- nition. pp. 18039–18048 (2024)

    2024

  23. [23]

    Commu- nications of the ACM65(1), 99–106 (2021)

    Mildenhall, B., Srinivasan, P.P., Tancik, M., Barron, J.T., Ramamoorthi, R., Ng, R.: Nerf: Representing scenes as neural radiance fields for view synthesis. Commu- nications of the ACM65(1), 99–106 (2021)

    2021

  24. [24]

    Journal of mathematical Psychology47(1), 90–100 (2003)

    Myung, I.J.: Tutorial on maximum likelihood estimation. Journal of mathematical Psychology47(1), 90–100 (2003)

    2003

  25. [25]

    National Institute of Standards and Technology: Gamma function (2025),https: //dlmf.nist.gov/5, nIST Digital Library of Mathematical Functions, Release 1.1.10

    2025

  26. [26]

    Oquab, M., Darcet, T., Moutakanni, T., Vo, H.V., Szafraniec, M., Khalidov, V., Fernandez,P.,Haziza,D.,Massa,F.,El-Nouby,A.,Howes,R.,Huang,P.Y.,Xu,H., Sharma, V., Li, S.W., Galuba, W., Rabbat, M., Assran, M., Ballas, N., Synnaeve, G., Misra, I., Jegou, H., Mairal, J., Labatut, P., Joulin, A., Bojanowski, P.: Dinov2: Learning robust visual features without ...

    2023

  27. [27]

    In: Proceedings of the AAAI Conference on Artificial Intelligence

    Qian, R., Cao, H., Deng, T., Yuan, S., Xie, L.: Splatssc: Decoupled depth-guided gaussian splatting for semantic scene completion. In: Proceedings of the AAAI Conference on Artificial Intelligence. vol. 40, pp. 8520–8528 (2026)

    2026

  28. [28]

    In: European conference on computer vision

    Silberman, N., Hoiem, D., Kohli, P., Fergus, R.: Indoor segmentation and support inference from rgbd images. In: European conference on computer vision. pp. 746–

  29. [29]

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

    Song, S., Yu, F., Zeng, A., Chang, A.X., Savva, M., Funkhouser, T.: Semantic scene completion from a single depth image. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 1746–1754 (2017)

    2017

  30. [30]

    Neurocomputing568, 127063 (2024)

    Su, J., Ahmed, M., Lu, Y., Pan, S., Bo, W., Liu, Y.: Roformer: Enhanced trans- former with rotary position embedding. Neurocomputing568, 127063 (2024)

    2024

  31. [31]

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

    Szymanowicz, S., Rupprecht, C., Vedaldi, A.: Splatter image: Ultra-fast single- view 3d reconstruction. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 10208–10217 (2024)

    2024

  32. [32]

    In: Interna- tional conference on machine learning

    Tan, M., Le, Q.: Efficientnetv2: Smaller models and faster training. In: Interna- tional conference on machine learning. pp. 10096–10106. PMLR (2021)

    2021

  33. [33]

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

    Tang, P., Wang, Z., Wang, G., Zheng, J., Ren, X., Feng, B., Ma, C.: Sparseocc: Rethinking sparse latent representation for vision-based semantic occupancy pre- diction. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2024)

    2024

  34. [34]

    Advances in neural information pro- cessing systems30(2017)

    Vaswani,A.,Shazeer,N.,Parmar,N.,Uszkoreit,J.,Jones,L.,Gomez,A.N.,Kaiser, Ł., Polosukhin, I.: Attention is all you need. Advances in neural information pro- cessing systems30(2017)

    2017

  35. [35]

    In: Proceedings of the 33rd ACM International Conference on Multimedia

    Wang, H., Wei, X., Zhang, X., Li, J., Bai, C., Li, Y., Lu, M., Zheng, W., Zhang, S.: Embodiedocc++: Boosting embodied 3d occupancy prediction with plane regular- ization and uncertainty sampler. In: Proceedings of the 33rd ACM International Conference on Multimedia. pp. 925–934 (2025)

    2025

  36. [36]

    In: Advances in Neural Information Processing Systems (2024)

    Wang, J., Liu, Z., Meng, Q., Yan, L., Wang, K., Yang, J., Liu, W., Hou, Q., Cheng, M.: Opus: occupancy prediction using a sparse set. In: Advances in Neural Information Processing Systems (2024)

    2024

  37. [37]

    In: Proceedings of the IEEE/CVF 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 occupancy perception. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 17850–17859 (2023)

    2023

  38. [38]

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

    Wimbauer, F., Yang, N., Rupprecht, C., Cremers, D.: Behind the scenes: Density fields for single view reconstruction. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 9076–9086 (2023)

    2023

  39. [39]

    In: Proceed- ings of the IEEE/CVF International Conference on Computer Vision (ICCV)

    Wu, Y., Zheng, W., Zuo, S., Huang, Y., Zhou, J., Lu, J.: Embodiedocc: Embodied 3d occupancy prediction for vision-based online scene understanding. In: Proceed- ings of the IEEE/CVF International Conference on Computer Vision (ICCV). pp. 26360–26370 (October 2025)

    2025

  40. [40]

    Advances in Neural Information Processing Systems37, 21875–21911 (2024)

    Yang, L., Kang, B., Huang, Z., Zhao, Z., Xu, X., Feng, J., Zhao, H.: Depth anything v2. Advances in Neural Information Processing Systems37, 21875–21911 (2024)

    2024

  41. [41]

    In: 2023 IEEE/CVF International Conference on Computer Vision (ICCV)

    Yao, J., Li, C., Sun, K., Cai, Y., Li, H., Ouyang, W., Li, H.: Ndc-scene: Boost monocular 3d semantic scene completion in normalized device coordinates space. In: 2023 IEEE/CVF International Conference on Computer Vision (ICCV). pp. 9421–9431. IEEE Computer Society (2023)

    2023

  42. [42]

    Quantum optics with giant atoms—the first five years

    Yu, H., Wang, Y., Chen, Y., Zhang, Z.: Monocular occupancy prediction for scal- able indoor scenes. In: Computer Vision – ECCV 2024: 18th European Conference, Milan, Italy, September 29–October 4, 2024, Proceedings, Part XXX. p. 38–54. 18 G. Chen, L. Huanget al. Springer-Verlag, Berlin, Heidelberg (2024).https://doi.org/10.1007/978- 3- 031-73404-5_3,https...

    work page doi:10.1007/978- 2024

  43. [43]

    Yu, Z., Shu, C., Deng, J., Lu, K., Liu, Z., Yu, J., Yang, D., Li, H., Chen, Y.: Flashocc: Fast and memory-efficient occupancy prediction via channel-to-height plugin (2023)

    2023

  44. [44]

    Advances in Neural Information Processing Systems37, 81489–81509 (2024)

    Zhang, G., Fan, L., He, C., Lei, Z., Zhang, Z., Zhang, L.: Voxel mamba: Group-free state space models for point cloud based 3d object detection. Advances in Neural Information Processing Systems37, 81489–81509 (2024)

    2024

  45. [45]

    In: European Conference on Computer Vision

    Zhang, K., Bi, S., Tan, H., Xiangli, Y., Zhao, N., Sunkavalli, K., Xu, Z.: Gs-lrm: Large reconstruction model for 3d gaussian splatting. In: European Conference on Computer Vision. pp. 1–19. Springer (2024)

    2024

  46. [46]

    In: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (June 2025), to appear

    Zhu, Z., Wang, S., Xie, J., Liu, J.J., Wang, J., Yang, J.: Voxelsplat: Dynamic gaus- sian splatting as an effective loss for occupancy and flow prediction. In: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (June 2025), to appear

    2025

  47. [47]

    In: Advances in Neural Information Processing Systems (2025)

    Zuo, S., Zheng, W., Han, X., Yang, L., Pan, Y., Lu, J.: Quadricformer: Scene as superquadrics for 3d semantic occupancy prediction. In: Advances in Neural Information Processing Systems (2025)

    2025

  48. [48]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Zuo, S., Zheng, W., Huang, Y., Zhou, J., Lu, J.: Gaussianworld: Gaussian world model for streaming 3d occupancy prediction. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 6772–6781 (2025) FLM for Indoor Occ 19 Supplementary Material A Discussion A.1 Limitations Bilinear Complexity of the FLM Loss. The computational cost of ...

    2025