Pith. sign in

REVIEW 2 major objections 5 minor 48 references

A training-free, multi-resolution geometric score that tells 3D part growth where to freeze can segment meshes as accurately as learned models—and more consistently across datasets.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

Hi-TOPS, a training-free meso-scale Flow-Freeze prior with TSDF-guided superquadric fitting, achieves competitive 3D part decomposition and the best mIoU on PartNet (55.87).

T0 review reviewed 2026-08-05 challenge →

load-bearing objection Hi-TOPS is a genuinely useful training-free mesh decomposition pipeline, but its headline quantitative claims rest on hyperparameters tuned on an evaluation subset, and its boundary score has a face-count dependence that should be tested before calling the prior 'stable'. the 2 major comments →

arxiv 2608.00767 v1 pith:ZCFHLVN7 submitted 2026-08-01 cs.GR cs.CV

Hi-TOPS: Hierarchical Topology-aware Scoring Prior for 3D Part Decomposition

Ruoyu Wu , Zhenhong Sun , Xiaoming Gong , Yuxin Xian , Zhi Wang , Yawen Chen , Huadong Mo , Daoyi Dong This is my paper
classification cs.GR cs.CV
keywords 3D part decompositionpart segmentationFlow-Freeze priortopology-aware scoringsuperquadric fittingTSDF-guided inflationmeso-scale voxelizationtraining-free shape analysis
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The reading

Hi-TOPS sets out to show that 3D part decomposition does not require semantic labels or large vision models: the shape's own geometry, read at the right scale, already says where parts separate. The paper identifies a structural-scale mismatch—global methods bridge over joints, local methods fragment under noise—and argues the decisive evidence lives at a middle scale. It therefore voxelizes the mesh at three resolutions (16, 32, 64), computes per-cell scores from curvature, normal variation, and sampling density, and splits the volume into Flow regions where primitives may grow and Freeze regions that stop growth at articulations and thin parts. Superquadrics are then inflated under a TSDF alignment objective with the Flow-Freeze map as a hard constraint, and the result is mapped back to mesh faces through curvature-aware edge primitives. The reported outcome is that this untrained pipeline beats primitive-based baselines, is competitive with trained models, and has the lowest cross-dataset mean-IoU variation (3.5) of any method tested, with its best result on PartNet (55.87).

Core claim

The central claim is that a hand-built, fixed-parameter geometric prior can stand in for learned semantics in part decomposition. On a voxelized mesh the paper defines a per-cell score S_B(i) = (λ1 μκ_i + λ2 σκ_i + λ3 Δκ_i + λ4 φ_norm,i) · sqrt(φ_den,i), computes it at resolutions 16, 32, and 64, thresholds the top 15% as Freeze, and fuses the three maps coarse-to-fine so that fine Freeze evidence survives while coarse Flow support is retained. The Flow-Freeze partition is then used as a hard constraint on superquadric inflation: seeds are placed first at the farthest interior points (bodies) and then at residual surface components (thin parts), growth is driven by minimizing a TSDF discrepa

What carries the argument

The Flow-Freeze prior is a scalar structural field on the active voxels of a mesh, S_B(i) = (λ1 μκ_i + λ2 σκ_i + λ3 Δκ_i + λ4 φ_norm,i) · sqrt(φ_den,i), whose top-α level set (α = 0.15) marks boundary-critical 'freeze' cells while the complement marks expandable 'flow' cells. The field is computed at resolutions {16, 32, 64} and fused by four octree rules, then used as a hard constraint in TSDF-guided superquadric inflation: it determines where body and surface seeds are placed, which expansions are accepted, and where growth stops so that articulation seams and thin attachments survive.

Load-bearing premise

The load-bearing premise is that a fixed score on coarse box-shaped cells—built from curvature, normal change, and sampling density, with the top 15 percent marked as freeze—reliably indicates where parts separate on arbitrary shapes; if a true boundary creates none of those cues, the method will miss it.

What would settle it

Feed Hi-TOPS a mesh whose ground-truth part boundary lies in a region of flat, low-curvature surface with no normal discontinuity and uniform sampling density, e.g., a cylinder whose cap-to-body seam has been smoothed so no geometric cue fires. If no voxel at the seam enters the top-15% Freeze set, the decomposition merges cap and body, falsifying the claim that the score is a reliable boundary proxy.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Because it needs no training data or 2D priors, the pipeline can be applied to any new mesh category without retraining or dataset-specific tuning.
  • The output is a set of editable superquadrics plus residual surface patches, so the same decomposition doubles as a compact abstraction for editing, animation, and controlled generation.
  • The structural-scale analysis implies that methods working only at global or only at local scales will keep bridging joints or fragmenting surfaces; boundary evidence at the meso scale is the missing ingredient.
  • On articulation-heavy, long-tailed benchmarks, the geometry-only prior is more consistent than trained models, indicating that distribution shift rather than representation quality is what limits those models.

Where Pith is reading between the lines

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

  • Beyond the paper: the Flow-Freeze field could be used as a free source of pseudo-labels to train or distill a fast network, since the prior itself requires no annotation.
  • Beyond the paper: because the paper reports that {32,64,128} beats {16,32,64} in accuracy at much higher cost, allocating fine voxels only where the score is locally uncertain would likely recover most of that gain at a fraction of the runtime.
  • Beyond the paper: the cue-to-boundary proxy could be stress-tested on deliberately nonuniform meshes and on shapes with smoothed, unarticulated seams; this would separate genuine geometric content from implementation choices.
  • Beyond the paper: replacing superquadrics with a more flexible primitive family, such as generalized cylinders or convex decompositions, while keeping the Flow-Freeze constraint should extend the method to highly non-convex parts.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 5 minor

Summary. The paper proposes Hi-TOPS, an unsupervised geometry-only method for 3D mesh part decomposition. It voxelizes a normalized mesh at three resolutions, computes per-voxel scores from mean/std/gradient of curvature, normal variation, and a density factor, then thresholds the top 15% as 'Freeze' regions. A rule-based octree fusion combines the three resolutions into a hierarchical Flow-Freeze prior. Superquadrics are seeded via EDT maxima ('Body') and residual-component centroids ('Surface'), inflated by TSDF-guided least-squares fits that must stay inside Flow voxels, and transferred back to the mesh via curvature-aware edge primitives with distance-weighted voting. On PartObjaverse-Tiny, PartNet-selected, and HY3D-Bench, Hi-TOPS reports mIoU 51.63, 55.87, and 47.22, outperforming primitive-based baselines, competitive with learned models, and showing the lowest cross-dataset mIoU standard deviation (3.5).

Significance. If the results hold, Hi-TOPS is a significant contribution: it demonstrates that a hand-designed meso-scale geometric prior can rival learned part segmentation without training data or 2D foundation models, and it provides a detailed, reproducible pipeline (pseudocode and hyperparameters in appendices). The PartNet result (+10.3 mIoU over PartField) is particularly striking. The main caveats are that the core score's tessellation sensitivity and the test-subset hyperparameter search are not yet resolved, so the strength of the quantitative claims is conditional. The paper ships detailed algorithm pseudocode, a full hyperparameter table, and component ablations, which are strengths.

major comments (2)
  1. [Eq. (3), App. B] The 'stable prior' claim is undermined because the score's density factor sqrt(phi_den) = sqrt(n_i) is face-count dependent. Since Nface=36 samples are placed per face, n_i is proportional to local triangle count, not surface area or geometric complexity. A 4x non-uniform subdivision of a patch quadruples n_i and doubles S_B in those voxels; because Freeze is the global top-15% quantile (Eqs. 4-5), remeshing can re-rank voxels and move Freeze labels from articulations to densely tessellated smooth regions. Since SQ fitting uses B_freeze as a hard constraint (Alg. A.2, flow-zone acceptance) and the edge-primitive assignment preserves freeze boundaries, this directly changes the output decomposition. The paper's own limitation (Sec. 5) says the cues 'may remain sensitive to tessellation' but no experiment quantifies this. Please add a remeshing-stability experiment (e.g., identical shape a
  2. [App. D.1-D.2, Table 4] The reported configuration is not a single truly dataset-agnostic setting: alpha, lambda, R, Nface, and q were fixed, but Appendix D.2 says the cue weights were selected by grid search on a fixed 36-shape subset of PartObjaverse-Tiny, and Table 4 uses the same subset to select the resolution hierarchy. No validation split is documented, and the final PartObjaverse-Tiny numbers in Table 1 are computed on the full 200-shape set that includes this subset. This makes part of the reported advantage on PartObjaverse-Tiny reflect selection on the test benchmark. Please either (a) introduce an explicit train/validation/test split, reporting main results on a held-out test set, or (b) report performance on all benchmarks under several reasonable hyperparameter choices to show the conclusions are insensitive. Note also that Table 4 shows {32,64,128} (mIoU 57.66) beats the chosen {16,32,64} (56.56)
minor comments (5)
  1. [Sec. 3.3 / App. E.2] Equation numbering is inconsistent: Eq. 8 in the main text is the SQ-to-mesh assignment, but App. E.2 (B.3) refers to Eq. 8 as the parameter-update acceptance rule. Renumber to avoid ambiguity.
  2. [Table 3] The row 'w/o TSDF Res=32' is placed under 'Geometric Cues', but removing TSDF is a fitting-stage ablation, not a geometric cue. Move it to the TSDF-related ablation group for clarity.
  3. [App. D.2] The sentence 'The table has been reorganized for clarity (same setting in Table 3)' appears to be an editorial leftover. Rewrite to describe the sensitivity table directly.
  4. [App. A / Table 1] The reported mIoU uses many-to-one greedy matching, which can be optimistic for over-segmented predictions. Please add a sensitivity check with Hungarian matching or report over/under-segmentation statistics.
  5. [Throughout] Minor typos: 'Hi-Tops' in App. A should be 'Hi-TOPS'; Fig. 2 caption has repetitive 'Normal Normal' formatting.

Circularity Check

1 steps flagged

Default weights and Freeze ratio were selected by grid search on a PartObjaverse-Tiny subset, so part of the reported PartObjaverse-Tiny advantage reflects test-set selection; the geometry-driven derivation itself is not circular.

specific steps
  1. fitted input called prediction [Appendix D.1 (Ablation Protocol), Appendix D.2 (Search for Cues' Weights), Table A.5; main results Table 1]
    "All ablations use a fixed 36-shape PartObjaverse-Tiny subset; ... Cue weights were determined via grid search. We found that uniform setting (1,1,1,1) yields stable performance across scenes and therefore adopt it as the default configuration."

    The paper presents a single dataset-agnostic setting (λ=(1,1,1,1), α=0.15) and then reports mIoU on PartObjaverse-Tiny (51.63) and cross-dataset std (3.5) as evidence of generalizability. But λ was selected by grid search on a 36-shape PartObjaverse-Tiny subset, and Table A.5 shows (1,1,1,1) is the best mIoU on that same subset. Table 3 similarly picks α=0.15 as the best of {5%,15%,25%} on the same subset. The PartObjaverse-Tiny column therefore is not an independent prediction from first principles; it partly reflects optimization against the evaluation set. The underlying score field (Eq. 3-8) never uses part labels, and the other two benchmarks are unaffected, so this is partial rather than total circularity.

full rationale

The central derivation is self-contained: S_B (Eq. 3) is a hand-specified geometric proxy over voxelized curvature/normal/density cues; Freeze is the top-α level set (Eq. 4-5); SQ fitting and assignment (Eq. 6-8, App. E) are geometry-only and never touch ground-truth labels. No load-bearing self-citation or uniqueness theorem is invoked; the citations to Light-SQ, DORA, and MPS are component borrowings, not justifications of the core claim. The main circular component is the selection of the default hyperparameters (λ, α) on a 36-shape subset of PartObjaverse-Tiny (App. D.1/D.2, Tables 3/A.5), followed by reporting PartObjaverse-Tiny mIoU and cross-dataset consistency in Table 1. That makes part of the headline advantage reflect test-set selection, but does not make the decomposition output equivalent to the fitted inputs. The limitation section also concedes that local cues may be sensitive to tessellation and dense details; this is a stability caveat, not a circularity.

Axiom & Free-Parameter Ledger

5 free parameters · 5 axioms · 0 invented entities

The pipeline introduces no physical entities; the Flow-Freeze field is an internal scoring construct. It rests on domain assumptions about geometric cues and meso-scale voxelization, plus five dominant hyperparameters fixed across datasets. No formal proof links the score to true part boundaries.

free parameters (5)
  • Freeze ratio alpha = 0.15
    Threshold in Eq. 4-5; ablation shows 5% and 25% are worse, 15% chosen from the same benchmark family.
  • Cue weights lambda = (1,1,1,1)
    Weights in Eq. 3; found via grid search on a 36-shape PartObjaverse-Tiny subset (App. D.2).
  • Resolution set R = {16,32,64}
    Chosen instead of {32,64,128}, which gives mIoU 57.66 vs 56.56; picked for lower runtime, not accuracy.
  • Dihedral cut quantile q = 90
    Edge-cut threshold for SQ-to-mesh assignment (App. E.3), fixed across datasets.
  • Per-face barycentric sample count Nface = 36
    Sampling density for cue aggregation, fixed default.
axioms (5)
  • ad hoc to paper Meso-scale voxel grids (16/32/64) capture the boundary evidence needed for decomposition.
    Sec. 3.1 asserts this after empirical observation; Table 4 shows 128 can improve accuracy, so the range is not derived.
  • domain assumption Curvature, normal inconsistency, and sampling density are sufficient intrinsic cues for part boundaries.
    Sec. 3.2 and Eq. 3 use these cues as a proxy; the paper provides no formal justification that they cover all articulation types.
  • domain assumption The top-alpha level set of the score separates boundary-critical from expandable regions.
    Eq. 4-5; this is a modeling choice.
  • domain assumption A superquadric fitted under TSDF alignment can represent body parts and thin attachments.
    Sec. 3.3; the limitations section admits very thin or non-convex parts may underfit.
  • standard math Standard voxelization and cotangent Laplacian computations are correct.
    Used without proof, standard in geometry processing.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Hi-TOPS: Hierarchical Topology-aware Scoring Prior for 3D Part Decomposition." pith.science (2026). https://pith.science/paper/ZCFHLVN7

@misc{pith2026260800767,
  author       = {Pith},
  title        = {Pith review of: Hi-TOPS: Hierarchical Topology-aware Scoring Prior for 3D Part Decomposition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZCFHLVN7}},
  note         = {Machine review of arXiv:2608.00767}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Accurate 3D part decomposition requires separating shapes into structurally meaningful components with precise boundaries while preserving articulation seams and thin attachments. Existing approaches often suffer from a structural-scale mismatch: geometric evidence for separation is most reliable at the meso scale, yet many pipelines operate either too globally to respect joints or too locally to remain robust to noise. We propose Hi-TOPS, a Hierarchical Topology-aware Scoring Prior that aggregates complementary intrinsic cues into a multi-resolution Flow-Freeze field. Flow regions provide expandable support for primitive coverage, while Freeze regions restrict growth near articulations and thin structures. A TSDF-guided body-surface superquadric fitter then captures dominant cores and residual surface structures, followed by SQ-to-mesh assignment for connected, boundary-aligned parts. Across diverse benchmarks, Hi-TOPS delivers stable, editable decompositions without semantic supervision or 2D foundation priors.

Figures

Figures reproduced from arXiv: 2608.00767 by Daoyi Dong, Huadong Mo, Ruoyu Wu, Xiaoming Gong, Yawen Chen, Yuxin Xian, Zhenhong Sun, Zhi Wang.

Figure 1
Figure 1. Figure 1: We showcase overall segmentation results in a composite scene, demonstrating the robustness of our [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Multi-scale voxelization and intrinsic cues. The leftmost panels show the input mesh and its wireframe complexity. Different voxel resolutions provide complementary evidence: coarse levels (𝑟 = 16/32) capture stable volumetric support, while finer levels (𝑟 = 64) reveal narrow joints and thin boundary details. The top panels visualize the three intrinsic geometric cues (Curvature, Normal, and Density), use… view at source ↗
Figure 3
Figure 3. Figure 3: Overview of the Hi-TOPS pipeline. The input mesh is voxelized at multiple meso-scales to compute intrinsic cues and construct a hierarchical Flow-Freeze prior. This prior guides body-surface SQ initialization, TSDF-guided SQ inflation, and SQ-to-mesh assignment, producing part decomposition while preserving local details. whether the SQ can safely grow nearby. We refine the SQ by mini￾mizing the TSDF discr… view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative comparison of 3D part segmentation on diverse meshes. We visualize segmentation results of primitive-based baselines, learning-based [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: More qualitative comparison case results. Our method demonstrates competitive segmentation performance with topologically faithful structure. [PITH_FULL_IMAGE:figures/full_fig_p010_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: From top to bottom, we show (a) the input mesh, (b) the hierarchical-resolution prior, (c) the superquadric (SQ) fitting results guided by the prior, and [PITH_FULL_IMAGE:figures/full_fig_p011_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: From left to right, we compare four settings from Table 4: (a) single-scale w TSDF / w/o Freeze, (b) single-scale w Freeze / w/o TSDF, (c) hierarchical w [PITH_FULL_IMAGE:figures/full_fig_p011_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Dataset-distribution analysis for PartSAM and PartField. We compare part-count, face-count, and category distributions across PartNet selected, [PITH_FULL_IMAGE:figures/full_fig_p011_8.png] view at source ↗

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 · 38 canonical work pages · 1 internal anchor

  1. [1]

    2025 , isbn =

    Ye, Jingwen and He, Yuze and Zhou, Yanning and Zhu, Yiqin and Xiao, Kaiwen and Liu, Yong-Jin and Yang, Wei and Han, Xiao , title =. 2025 , isbn =. doi:10.1145/3721238.3730732 , booktitle =

    work page arXiv 2025

  2. [2]

    2025 , isbn =

    Wang, Yuhan and Chen, Weikai and Hu, Zeyu and Zhang, Runze and Yin, Yingda and Wu, Ruoyu and Luo, Keyang and Qian, Shengju and Ma, Yiyan and Li, Hongyi and Gao, Yuan and Zhou, Yuhuan and Luo, Hao and Wang, Wan and Shen, Xiaobin and Li, Zhaowei and Zhu, Kuixin and Hong, Chuanlang and Wang, Yueyue and Feng, Lijie and Wang, Xin and Loy, Chen Change , title =...

    work page arXiv 2025

  3. [3]

    Proceedings IEEE Conf

    Marching-Primitives: Shape Abstraction from Signed Distance Function , author =. Proceedings IEEE Conf. on Computer Vision and Pattern Recognition (CVPR) , year =

    work page

  4. [4]

    , title =

    Liu, Weixiao and Wu, Yuwei and Ruan, Sipu and Chirikjian, Gregory S. , title =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , month =. 2022 , pages =

    work page 2022

  5. [5]

    and Efros, Alexei A

    Tulsiani, Shubham and Su, Hao and Guibas, Leonidas J. and Efros, Alexei A. and Malik, Jitendra , booktitle=. Learning Shape Abstractions by Assembling Volumetric Primitives , year=

    work page

  6. [6]

    Robust Shape Fitting for 3D Scene Abstraction , year=

    Kluger, Florian and Brachmann, Eric and Yang, Michael Ying and Rosenhahn, Bodo , journal=. Robust Shape Fitting for 3D Scene Abstraction , year=

    work page

  7. [7]

    , title =

    Hilaga, Masaki and Shinagawa, Yoshihisa and Kohmura, Taku and Kunii, Tosiyasu L. , title =. Proceedings of the 28th Annual Conference on Computer Graphics and Interactive Techniques , pages =. 2001 , isbn =. doi:10.1145/383259.383282 , abstract =

    work page arXiv 2001

  8. [8]

    Discrete Comput

    Edelsbrunner and Letscher and Zomorodian , title =. Discrete Comput. Geom. , month = nov, pages =. 2002 , issue_date =. doi:10.1007/s00454-002-2885-2 , abstract =

    doi:10.1007/s00454-002-2885-2 2002

  9. [9]

    IEEE Transactions on Visualization and Computer Graphics , month = nov, pages =

    Gyulassy, Attila and Natarajan, Vijay and Pascucci, Valerio and Hamann, Bernd , title =. IEEE Transactions on Visualization and Computer Graphics , month = nov, pages =. 2007 , issue_date =. doi:10.1109/TVCG.2007.70552 , abstract =

    work page arXiv 2007

  10. [10]

    2025 , eprint=

    P3-SAM: Native 3D Part Segmentation , author=. 2025 , eprint=

    work page 2025

  11. [11]

    Changfeng Ma and Yang Li and Xinhao Yan and Jiachen Xu and Yunhan Yang and Chunshi Wang and Zibo Zhao and Yanwen Guo and Zhuo Chen and Chunchao Guo , booktitle=. P\. 2025 , url=

    work page 2025

  12. [12]

    Zhe Zhu and Le Wan and Rui Xu and Yiheng Zhang and Honghua Chen and Zhiyang Dou and Cheng Lin and Yuan Liu and Mingqiang Wei , journal=

    work page

  13. [13]

    Qi and Su, Hao and Kaichun, Mo and Guibas, Leonidas J

    Charles, R. Qi and Su, Hao and Kaichun, Mo and Guibas, Leonidas J. , booktitle=. PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation , year=

    work page

  14. [14]

    SGPN: Similarity Group Proposal Network for 3D Point Cloud Instance Segmentation , year=

    Wang, Weiyue and Yu, Ronald and Huang, Qiangui and Neumann, Ulrich , booktitle=. SGPN: Similarity Group Proposal Network for 3D Point Cloud Instance Segmentation , year=

    work page

  15. [15]

    2025 , eprint=

    Segment Any Mesh , author=. 2025 , eprint=

    work page 2025

  16. [16]

    IEEE/CVF International Conference on Computer Vision (CVPR) , year =

    PartDistill: 3D Shape Part Segmentation by Vision-Language Model Distillation , author =. IEEE/CVF International Conference on Computer Vision (CVPR) , year =

    work page

  17. [17]

    S2AM3D: Scale-controllable Part Segmentation of 3D Point Clouds

    Su, Han and Huang, Tianyu and Wan, Zichen and Wu, Xiaohe and Zuo, Wangmeng , title =. arXiv preprint arXiv:2512.00995 , year =

    work page internal anchor Pith review Pith/arXiv arXiv

  18. [18]

    IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) , year =

    Garosi, Marco and Tedoldi, Riccardo and Boscaini, Davide and Mancini, Massimiliano and Sebe, Nicu and Poiesi, Fabio , title =. IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) , year =

    work page

  19. [19]

    PartSLIP: Low-Shot Part Segmentation for 3D Point Clouds via Pretrained Image-Language Models , year=

    Liu, Minghua and Zhu, Yinhao and Cai, Hong and Han, Shizhong and Ling, Zhan and Porikli, Fatih and Su, Hao , booktitle=. PartSLIP: Low-Shot Part Segmentation for 3D Point Clouds via Pretrained Image-Language Models , year=

    work page

  20. [20]

    Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , month =

    Liu, Minghua and Uy, Mikaela Angelina and Xiang, Donglai and Su, Hao and Fidler, Sanja and Sharp, Nicholas and Gao, Jun , title =. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , month =. 2025 , pages =

    2025

  21. [21]

    MirageRoom: 3D Scene Segmentation with 2D Pre-Trained Models by Mirage Projection , year=

    Sun, Haowen and Duan, Yueqi and Yan, Juncheng and Liu, Yifan and Lu, Jiwen , booktitle=. MirageRoom: 3D Scene Segmentation with 2D Pre-Trained Models by Mirage Projection , year=

    work page

  22. [22]

    2024 , eprint=

    SAMPart3D: Segment Any Part in 3D Objects , author=. 2024 , eprint=

    work page 2024

  23. [23]

    Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , month =

    Ma, Ziqi and Yue, Yisong and Gkioxari, Georgia , title =. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , month =. 2025 , pages =

    work page 2025

  24. [24]

    2023 , eprint=

    PartSLIP++: Enhancing Low-Shot 3D Part Segmentation via Multi-View Instance Segmentation and Maximum Likelihood Estimation , author=. 2023 , eprint=

    work page 2023

  25. [25]

    Yuchen Zhou and Jiayuan Gu and Tung Yen Chiang and Fanbo Xiang and Hao Su , booktitle=. Point-. 2025 , url=

    work page 2025

  26. [26]

    arXiv:2304.02643 , year=

    Segment Anything , author=. arXiv:2304.02643 , year=

    Pith/arXiv arXiv

  27. [27]

    arXiv:2304.07193 , year=

    DINOv2: Learning Robust Visual Features without Supervision , author=. arXiv:2304.07193 , year=

    Pith/arXiv arXiv

  28. [28]

    CVPR , year=

    Point Transformer V3: Simpler, Faster, Stronger , author=. CVPR , year=

    work page

  29. [29]

    CVPR , year=

    Sonata: Self-Supervised Learning of Reliable Point Representations , author=. CVPR , year=

  30. [30]

    2022 , booktitle=

    Grounded Language-Image Pre-training , author=. 2022 , booktitle=

    2022

  31. [31]

    arXiv preprint arXiv:2206.05836 , year=

    GLIPv2: Unifying Localization and Vision-Language Understanding , author=. arXiv preprint arXiv:2206.05836 , year=

    Pith/arXiv arXiv

  32. [32]

    2014 , isbn =

    Girshick, Ross and Donahue, Jeff and Darrell, Trevor and Malik, Jitendra , title =. 2014 , isbn =. doi:10.1109/CVPR.2014.81 , booktitle =

    doi:10.1109/cvpr.2014.81 2014

  33. [33]

    ACM Transactions on Graphics (TOG) , volume=

    Approximate convex decomposition for 3d meshes with collision-aware concavity and tree search , author=. ACM Transactions on Graphics (TOG) , volume=. 2022 , publisher=

    work page 2022

  34. [34]

    and Whitaker, R.T

    Mangan, A.P. and Whitaker, R.T. , journal=. Partitioning 3D surface meshes using watershed segmentation , year=

    work page

  35. [35]

    ACM Trans

    Cohen-Steiner, David and Alliez, Pierre and Desbrun, Mathieu , title =. ACM Trans. Graph. , month = aug, pages =. 2004 , issue_date =. doi:10.1145/1015706.1015817 , abstract =

    work page arXiv 2004

  36. [36]

    Proceedings of the 5th International Conference on Advances in Geometric Modeling and Processing , pages =

    Zhang, Xi and Li, Guiqing and Xiong, Yunhui and He, Fenghua , title =. Proceedings of the 5th International Conference on Advances in Geometric Modeling and Processing , pages =. 2008 , isbn =

    work page 2008

  37. [37]

    The Visual Computer , volume =

    Mesh segmentation driven by Gaussian curvature , author =. The Visual Computer , volume =. 2005 , publisher =. doi:10.1007/s00371-005-0319-x , url =

    work page doi:10.1007/s00371-005-0319-x 2005

  38. [38]

    and Yi, Li and Tripathi, Subarna and Guibas, Leonidas J

    Mo, Kaichun and Zhu, Shilin and Chang, Angel X. and Yi, Li and Tripathi, Subarna and Guibas, Leonidas J. and Su, Hao , title =. The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , month =

    work page

  39. [39]

    2026 , eprint=

    HY3D-Bench: Generation of 3D Assets , author=. 2026 , eprint=

    work page 2026

  40. [40]

    PlanarRecon: Realtime 3D Plane Detection and Reconstruction from Posed Monocular Videos , year=

    Xie, Yiming and Gadelha, Matheus and Yang, Fengting and Zhou, Xiaowei and Jiang, Huaizu , booktitle=. PlanarRecon: Realtime 3D Plane Detection and Reconstruction from Posed Monocular Videos , year=

    work page

  41. [41]

    Proceedings of the 38th International Conference on Neural Information Processing Systems , articleno =

    He, Yuze and Zhao, Wang and Liu, Shaohui and Hu, Yubin and Bai, Yushi and Wen, Yu-Hui and Liu, Yong-Jin , title =. Proceedings of the 38th International Conference on Neural Information Processing Systems , articleno =. 2024 , isbn =

    work page 2024

  42. [42]

    arXiv preprint arXiv:1801.09847 , year=

    Open3D: A modern library for 3D data processing , author=. arXiv preprint arXiv:1801.09847 , year=

    Pith/arXiv arXiv

  43. [43]

    arXiv preprint arXiv:2509.08643 , year=

    X-Part: high fidelity and structure coherent shape decomposition , author=. arXiv preprint arXiv:2509.08643 , year=

    arXiv

  44. [44]

    2025 , eprint=

    OmniPart: Part-Aware 3D Generation with Semantic Decoupling and Structural Cohesion , author=. 2025 , eprint=

    work page 2025

  45. [45]

    2025 , eprint=

    HoloPart: Generative 3D Part Amodal Segmentation , author=. 2025 , eprint=

    work page 2025

  46. [46]

    BANG: Dividing 3D Assets via Generative Exploded Dynamics , volume=

    Zhang, Longwen and Zhang, Qixuan and Jiang, Haoran and Bai, Yinuo and Yang, Wei and Xu, Lan and Yu, Jingyi , year=. BANG: Dividing 3D Assets via Generative Exploded Dynamics , volume=. ACM Transactions on Graphics , publisher=. doi:10.1145/3730840 , number=

    work page doi:10.1145/3730840

  47. [47]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Dora: Sampling and benchmarking for 3d shape variational auto-encoders , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    work page

  48. [48]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Richdreamer: A generalizable normal-depth diffusion model for detail richness in text-to-3d , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    work page

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.