Pith. sign in

REVIEW 2 major objections 5 minor 75 references

SGMatch: Semantic-Guided Non-Rigid Shape Matching with Flow Regularization

T0 review · 2 major / 5 minor · reviewed 2026-07-14 · grok-4.5

Pith's one-line read SGMatch recovers accurate point-to-point maps on non-rigid 3D shapes by fusing vision-model semantics with flow-matching smoothness, improving most under non-isometry and topological noise.

desk verdict Solid functional-map engineering paper: SGLCA + CFM give real gains on SMAL/TOPKIDS; ablations mostly clean, joint contribution only partially isolated. read the letter →

arxiv 2603.12937 v2 pith:GW52JTWF submitted 2026-03-13 cs.CV

classification cs.CV
keywords ShapeMatchingFunctionalMapsFlowNon-RigidCorrespondenceSemanticFeatures3DAnalysisConditional
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

Finding which points on one deformable 3D surface match which points on another remains hard when shapes stretch differently, meshes contain topological defects, or geometry alone cannot tell left from right. This paper claims those failures of functional-map pipelines can be reduced by two complementary ingredients. First, multi-view semantic features from a frozen vision foundation model are lifted onto the mesh and fused into geometric descriptors through a gated local cross-attention module that respects neighborhood structure. Second, conditional flow matching is adapted as a regularizer that trains a time-conditioned velocity field so feature transport between corresponding points stays spatially coherent. Trained without ground-truth maps, the resulting framework stays competitive on ordinary near-isometric human data while cutting error on non-isometric animal shapes, cross-category humans, and topologically noisy scans. A sympathetic reader cares because texture transfer, pose transfer, and statistical shape analysis all depend on dense maps that remain reliable precisely when pure geometry breaks down.

What carries the argument

Semantic-Guided Local Cross-Attention (SGLCA), which gates geometric descriptors with projected semantic features and restricts attention to local mesh neighborhoods, together with a conditional flow matching objective that supervises a time-varying velocity field along linear feature-transport paths after spectral heat diffusion.

What would settle it

Train and evaluate the full model versus its geometric-only ablation on a new non-isometric shape family whose appearance lies far outside the foundation model's pretraining distribution; if the semantic-plus-flow gains vanish while the same model still wins on SMAL and TOPKIDS, the claim that the lifted semantics are reliable anchors is falsified.

Watch

Extended reading notes

Core claim

SGMatch establishes that coupling 3D-lifted semantic cues from vision foundation models with trajectory-level feature transport regularization produces competitive near-isometric matching and consistent gains under non-isometric deformations and topological noise, reporting the lowest mean geodesic errors among compared unsupervised methods on SMAL and TOPKIDS while also lowering conformal distortion.

Load-bearing premise

The method assumes that multi-view features lifted from a frozen vision foundation model supply stable semantic anchors that correctly line up corresponding body parts across different shapes and species without category-specific training.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 5 minor

Summary. The paper introduces SGMatch, an unsupervised deep functional-map pipeline for non-rigid 3D shape matching. It fuses multi-view-lifted DINOv2 semantic features into geometric descriptors via a Semantic-Guided Local Cross-Attention (SGLCA) module that uses channel gating and 1-ring neighborhood attention, then regularizes the recovered soft point-wise maps by supervising a time-conditioned velocity field under a conditional flow-matching (CFM) objective after spectral heat diffusion. The method is evaluated on FAUST/SCAPE/SHREC’19 (near-isometric), SMAL and DT4D-H (non-isometric), and TOPKIDS (topological noise), reporting competitive geodesic errors on near-isometric data and state-of-the-art numbers on the harder regimes (SMAL 2.5, TOPKIDS 3.3), supported by PCK/AUC curves, conformal-distortion plots, multi-seed statistics, runtime/memory figures, and component ablations.

Significance. If the reported gains hold under independent re-implementation, the work supplies a practical, modular way to inject frozen vision-foundation-model semantics into the classical functional-map pipeline while adding a trajectory-level smoothness prior. This directly addresses two long-standing failure modes—symmetry ambiguity and local spatial inconsistency under non-isometry or topological noise—without requiring part labels or category-specific fine-tuning. The multi-seed stability numbers, conformal-distortion analysis, and extensive appendix ablations (fusion variants, sampling, neighborhood size, diffusion time) constitute reproducible evidence that strengthens the empirical claim beyond typical conference tables.

major comments (2)
  1. [§5.5, Table 2] The central performance claim on non-isometric data rests on the joint contribution of SGLCA semantics and CFM, yet Table 2 (SMAL) does not furnish a complete factorial. Row I removes semantics while retaining heat diffusion + CFM (error 3.2); row V removes both heat diffusion and CFM while keeping semantics (error 2.7). Consequently it remains impossible to quantify how much of the 0.8-point improvement over HybridFMap (3.3) is carried by the foundation-model anchors versus the flow regularizer alone. A 2 imes2 (semantics on/off imes CFM on/off, heat fixed) on the same split is required to make the attribution load-bearing.
  2. [§4.4, Eq. (16)] The CFM regularizer (Eqs. 13–16) is trained on soft maps that are themselves being optimized; early-training soft maps can be highly inaccurate. While the Charbonnier loss and importance sampling mitigate outliers, no diagnostic is provided that the learned velocity field does not reinforce incorrect local matches. A simple plot of geodesic error (or conformal distortion) versus training epoch, with and without CFM, would confirm that the trajectory constraint improves rather than locks in early errors.
minor comments (5)
  1. [Abstract / §1] The abstract and introduction repeatedly claim “consistent improvements under non-isometric deformations and topological noise.” On DT4D-H inter-class the absolute gain over HybridFMap is only 0.1; a more precise phrasing (“best or second-best, with largest relative gains on SMAL and TOPKIDS”) would avoid overstatement.
  2. [§4.2 / App. E.1] Neighborhood size k is fixed at 32 with self-loop padding (App. E.1). A short remark on how this interacts with meshes of highly varying valence would help readers port the module.
  3. [§4.1] Semantic features are obtained by multi-view rendering + frozen DINOv2 + averaging (Eqs. 3–4). The text never states whether the same camera set is used at test time or whether the features are cached; a one-sentence clarification would aid reproducibility.
  4. [Fig. 1] Figure 1 (left) caption asserts that semantic features “resolve ambiguity,” yet the qualitative example is only visual; a quantitative left/right symmetry error (common in the functional-map literature) would make the claim sharper.
  5. [Throughout] Several concatenated words appear in the arXiv text (e.g., “Establishingaccurate,” “non-isometricdeformations”). These are presumably PDF extraction artifacts, but a clean re-compilation would improve readability.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: empirical unsupervised learning method evaluated on held-out public benchmarks with standard non-forcing losses and ablations.

full rationale

SGMatch is a deep functional-maps pipeline that fuses multi-view DINOv2 semantic features via a local cross-attention module and adds a conditional-flow-matching regularizer on soft correspondences. The training objective (L_spectral + λ_cfm L_cfm) comprises ordinary structural/bijectivity/orthogonality terms, a coupling residual between functional and pointwise maps, and a Charbonnier velocity-matching loss along linear feature paths; none of these algebraically define or force the reported mean geodesic errors. All quantitative claims are obtained by training on designated train splits and measuring geodesic error / PCK / conformal distortion on completely held-out test shapes from FAUST, SCAPE, SHREC’19, SMAL, DT4D-H and TOPKIDS. Ablations (Table 2) and parameter sweeps isolate components without circular re-use of the evaluation metric. Self-citations (e.g., authors’ prior DCMATCH or multi-shape papers) appear only in related-work lists and are not invoked as uniqueness theorems or load-bearing premises. Consequently the derivation chain never reduces a claimed prediction to its own inputs by construction.

Assumptions & free parameters 5 free parameters · 3 assumptions · 2 invented entities

The method rests on standard functional-map and flow-matching mathematics plus the empirical premise that frozen DINOv2 features remain useful after multi-view lifting. Free parameters are the usual training hyper-parameters and a few design constants (neighborhood size, diffusion time, CFM weight). No new physical entities are postulated; SGLCA and the CFM regularizer are engineering constructs whose value is measured by the reported errors.

free parameters (5)
  • λ_cfm = 100
    Weight of the conditional flow-matching term set to 100; controls the strength of the smoothness regularizer relative to spectral losses.
  • diffusion time τ = 10^{-2} or 10^{-4}
    Spectral heat-diffusion scale chosen per regime (10^{-2} near-isometric/topology, 10^{-4} non-isometric); directly affects feature endpoints for CFM.
  • confidence concentration α = 2
    Temperature of importance sampling weights in the CFM objective; set to 2.
  • neighborhood size k = 32
    Fixed 1-ring padded size for local cross-attention; set to 32.
  • temperature τ_T = 0.07
    Softmax temperature for soft point-wise maps; set to 0.07.
assumptions (3)
  • domain assumption Laplace–Beltrami eigenfunctions truncated to k dimensions form a sufficient basis for functional maps under the deformations of interest.
    Standard functional-map premise (Sec. 3.1); known to degrade under strong non-isometry and topology noise—the very regimes the paper targets.
  • domain assumption Multi-view-averaged DINOv2 features back-projected onto the mesh supply semantically consistent vertex descriptors across non-isometric instances.
    Core premise of Sec. 4.1; inherited from Diff3F-style lifting and left largely unproven for arbitrary new categories.
  • ad hoc to paper Linear interpolation paths in feature space plus a learned velocity field yield spatially coherent correspondences when supervised by soft maps.
    Adaptation of CFM (Sec. 4.4) to the correspondence setting; justified empirically by ablation and conformal-distortion plots.
invented entities (2)
  • Semantic-Guided Local Cross-Attention (SGLCA) module
    purpose: Fuse semantic and geometric features with channel gating and neighborhood-restricted attention while preserving local continuity.
    New architectural block; value demonstrated only by the paper’s own ablations.
  • Conditional flow matching regularization for soft point-wise maps
    purpose: Encourage trajectory-level spatial smoothness of recovered correspondences via a time-conditioned velocity field.
    Task-specific adaptation of CFM; independent evidence limited to the reported smoothness metrics.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SGMatch: Semantic-Guided Non-Rigid Shape Matching with Flow Regularization." pith.science (2026). https://pith.science/paper/GW52JTWF

@misc{pith2026260312937,
  author       = {Pith},
  title        = {Pith review of: SGMatch: Semantic-Guided Non-Rigid Shape Matching with Flow Regularization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GW52JTWF}},
  note         = {Machine review of arXiv:2603.12937}
}
read the original abstract

Establishing accurate point-to-point correspondences between non-rigid 3D shapes remains a critical challenge, particularly under non-isometric deformations and topological noise. Existing functional map pipelines suffer from ambiguities that geometric descriptors alone cannot resolve, and spatial inconsistencies inherent in the projection of truncated spectral bases to dense pointwise correspondences. In this paper, we introduce SGMatch, a learning-based framework that couples 3D-lifted semantic cues with trajectory-level feature transport regularization. Specifically, we design a Semantic-Guided Local Cross-Attention module that integrates semantic features from vision foundation models into geometric descriptors while preserving local structural continuity. Furthermore, we adapt conditional flow matching as a time-conditioned feature transport regularizer that promotes spatially coherent point-wise recovery. Experimental results on multiple benchmarks demonstrate that SGMatch achieves competitive performance across near-isometric settings and consistent improvements under non-isometric deformations and topological noise.

Figures

Figures reproduced from arXiv: 2603.12937 by the authors.

Figure 1
Figure 1. (Left): Colormap transfer on the SHREC’19 dataset demonstrates that in￾corporating semantic features resolves ambiguity and yields globally consistent corre￾spondences. (Right): Vertex transfer on the SMAL dataset shows that the proposed conditional flow matching regularization promotes spatially smooth correspondences. robust zero-shot descriptors. However, integrating these signals into functional map pipelines is… view at source ↗
Figure 2
Figure 2. Overview of SGMatch. Given a pair of shapes X and Y, we extract geo￾metric features F geo X , F geo Y and semantic features F sem X , F sem Y , which are subsequently fused via the proposed SGLCA module. The resulting fused representations F fuse X and F fuse Y are then used to estimate functional maps CX Y and to recover dense point-wise correspondences ΠYX . In parallel, spectral heat diffusion followed by conditi… view at source ↗
Figure 3
Figure 3. Left: Near-isometric matching and cross-dataset generalisation on FAUST, SCAPE, and SHREC’19. Best results are highlighted. Right: Qualitative results on the challenging SHREC’19 dataset. Results We adopt the mean geodesic error as the evaluation metric [32]. Quan￾titative and qualitative results in [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: PCK curves and AUC values. Left: Non-isometric matching on SMAL and DT4D-H. Right: Matching with topological noise on TOPKIDS. Our approach achieves strong performance in both settings, outperforming existing methods. competitive intra-class results and the best perfor…
Figure 5
Figure 5. Figure 5: Qualitative Results on SMAL and DT4D-H. Comparison of our method against DeepFAFM and HybridFMap, via texture transfer. Geo.Err TOPKIDS Axiomatic Methods ZoomOut [45] 33.7 Smooth Shells [21] 10.8 DiscreteOp [54] 35.5 Unsupervised Methods Deep Shell [23] 13.7 AttnFMaps …
Figure 6
Figure 6. Figure 6: Left: Quantitative comparison with state-of-the-art methods on TOPKIDS under topological noise. Right: Qualitative comparison on TOPKIDS; our method produces more accurate, coherent correspondences than existing approaches. map-based methods, which rely on stable Lapla…
Figure 7
Figure 7. Figure 7: Conformal-distortion-based smoothness comparison. [PITH_FULL_IMAGE:figures/full_fig_p013_7.png]
Figure 8
Figure 8. Figure 8: Feature similarity heatmaps on SMAL cross-species pairs. [PITH_FULL_IMAGE:figures/full_fig_p022_8.png]
Figure 9
Figure 9. Figure 9: Effect of CFM regularization on correspondence smoothness for a [PITH_FULL_IMAGE:figures/full_fig_p023_9.png]
Figure 10
Figure 10. Figure 10: Additional qualitative results of our method on SHREC’19 [PITH_FULL_IMAGE:figures/full_fig_p026_10.png]
Figure 11
Figure 11. Figure 11: Additional qualitative results of our method on SMAL. Source [PITH_FULL_IMAGE:figures/full_fig_p027_11.png]
Figure 12
Figure 12. Figure 12: Additional qualitative results of our method on DT4D-H inter class. Source [PITH_FULL_IMAGE:figures/full_fig_p027_12.png]
Figure 13
Figure 13. Figure 13: Additional qualitative results of our method on TOPKIDS [PITH_FULL_IMAGE:figures/full_fig_p027_13.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

75 extracted references · 6 linked inside Pith

  1. [1]

    In: SIGGRAPH Asia 2023 Conference Papers

    Abdelreheem, A., Eldesokey, A., Ovsjanikov, M., Wonka, P.: Zero-shot 3d shape correspondence. In: SIGGRAPH Asia 2023 Conference Papers. pp. 1–11 (2023)

  2. [2]

    In: ACM Siggraph 2005 Papers, pp

    Anguelov, D., Srinivasan, P., Koller, D., Thrun, S., Rodgers, J., Davis, J.: Scape: shape completion and animation of people. In: ACM Siggraph 2005 Papers, pp. 408–416 (2005)

  3. [3]

    In: ICCV

    Aubry, M., Schlickewei, U., Cremers, D.: The wave kernel signature: A quantum mechanical approach to shape analysis. In: ICCV. pp. 1626–1633. IEEE (2011)

  4. [4]

    ACM Transactions on Multimedia Computing, Communications and Applications20(3), 1–24 (2023)

    Baldrati, A., Bertini, M., Uricchio, T., Del Bimbo, A.: Composed image retrieval using contrastive learning and task-oriented clip-based features. ACM Transactions on Multimedia Computing, Communications and Applications20(3), 1–24 (2023)

  5. [5]

    In: CVPR

    Bastian, L., Xie, Y., Navab, N., Lähner, Z.: Hybrid functional maps for crease- aware non-isometric shape matching. In: CVPR. pp. 3313–3323 (2024)

  6. [6]

    In: ICML

    Behmanesh, M., Krahn, M., Ovsjanikov, M.: Tide: Time derivative diffusion for deep learning on graphs. In: ICML. pp. 2015–2030. PMLR (2023)

  7. [7]

    In: CVPR

    Bogo, F., Romero, J., Loper, M., Black, M.J.: Faust: Dataset and evaluation for 3d mesh registration. In: CVPR. pp. 3794–3801 (2014)

  8. [8]

    In: CVPR

    Bronstein, M.M., Kokkinos, I.: Scale-invariant heat kernel signatures for non-rigid shape recognition. In: CVPR. pp. 1704–1711. IEEE (2010)

Show all 75 references
  1. [9]

    In: ECCV

    Cao, D., Bernard, F.: Unsupervised deep multi-shape matching. In: ECCV. pp. 55–71. Springer (2022)

  2. [10]

    In: ECCV

    Cao, D., Lähner, Z., Bernard, F.: Synchronous diffusion for unsupervised smooth non-rigid 3d shape matching. In: ECCV. pp. 262–281. Springer (2024)

  3. [11]

    ACM TOG42(4), 1–15 (2023)

    Cao, D., Roetzer, P., Bernard, F.: Unsupervised learning of robust spectral shape matching. ACM TOG42(4), 1–15 (2023)

  4. [12]

    In: CVPR

    Caron, M., Touvron, H., Misra, I., Jégou, H., Mairal, J., Bojanowski, P., Joulin, A.: Emerging properties in self-supervised vision transformers. In: CVPR. pp. 9650– 9660 (2021)

  5. [13]

    In: CVPR

    Chen, Z., Jiang, P., Huang, R.: Dv-matcher: Deformation-based non-rigid point cloud matching guided by pre-trained visual features. In: CVPR. pp. 27264–27274 (2025)

  6. [14]

    ACM TOG24(2), 289–310 (2005)

    Dinh, H.Q., Yezzi, A., Turk, G.: Texture transfer during shape transformation. ACM TOG24(2), 289–310 (2005)

  7. [15]

    In: CVPR

    Donati, N., Corman, E., Ovsjanikov, M.: Deep orientation-aware functional maps: Tackling symmetry issues in shape matching. In: CVPR. pp. 742–751 (2022)

  8. [16]

    In: CVPR

    Donati, N., Sharma, A., Ovsjanikov, M.: Deep geometric functional maps: Robust feature learning for shape correspondence. In: CVPR. pp. 8592–8601 (2020)

  9. [17]

    In: CVPR

    Dutt, N.S., Muralikrishnan, S., Mitra, N.J.: Diffusion 3d features (diff3f): Decorat- ing untextured shapes with distilled semantic features. In: CVPR. pp. 4494–4504 (2024)

  10. [18]

    ACM TOG39(5), 1–38 (2020)

    Egger,B.,Smith,W.A.,Tewari,A.,Wuhrer,S.,Zollhoefer,M.,Beeler,T.,Bernard, F., Bolkart, T., Kortylewski, A., Romdhani, S., et al.: 3d morphable face mod- els—past, present, and future. ACM TOG39(5), 1–38 (2020)

  11. [19]

    In: Computer Graphics Forum

    Ehm, V., Amrani, N.E., Xie, Y., Bastian, L., Gao, M., Wang, W., Sang, L., Cao, D., Weißberg, T., Lähner, Z., et al.: Beyond complete shapes: A benchmark for quantitative evaluation of 3d shape surface matching algorithms. In: Computer Graphics Forum. vol. 44, p. e70186. Wiley ...

  12. [20]

    arXiv preprint arXiv:2309.05013 (2023) 16 Tianwei Ye et al

    Ehm, V., Roetzer, P., Eisenberger, M., Gao, M., Bernard, F., Cremers, D.: Ge- ometrically consistent partial shape matching. arXiv preprint arXiv:2309.05013 (2023) 16 Tianwei Ye et al

  13. [21]

    In: CVPR

    Eisenberger, M., Lahner, Z., Cremers, D.: Smooth shells: Multi-scale shape regis- tration with functional maps. In: CVPR. pp. 12265–12274 (2020)

  14. [22]

    In: CVPR

    Eisenberger, M., Novotny, D., Kerchenbaum, G., Labatut, P., Neverova, N., Cre- mers, D., Vedaldi, A.: Neuromorph: Unsupervised shape interpolation and corre- spondence in one go. In: CVPR. pp. 7473–7483 (2021)

  15. [23]

    NeurIPS33, 10491–10502 (2020)

    Eisenberger, M., Toker, A., Leal-Taixé, L., Cremers, D.: Deep shells: Unsupervised shape correspondence with optimal transport. NeurIPS33, 10491–10502 (2020)

  16. [24]

    In: 2016 Fourth International Conference on 3D Vision (3DV)

    Eynard, D., Rodola, E., Glashoff, K., Bronstein, M.M.: Coupled functional maps. In: 2016 Fourth International Conference on 3D Vision (3DV). pp. 399–407. IEEE (2016)

  17. [25]

    In: CVPR

    Gao, M., Lahner, Z., Thunberg, J., Cremers, D., Bernard, F.: Isometric multi-shape matching. In: CVPR. pp. 14183–14193 (2021)

  18. [26]

    In: ECCV

    Groueix, T., Fisher, M., Kim, V.G., Russell, B.C., Aubry, M.: 3d-coded: 3d corre- spondences by deep deformation. In: ECCV. pp. 230–246 (2018)

  19. [27]

    arXiv preprint arXiv:2506.02070 (2025)

    Holderrieth, P., Erives, E.: An introduction to flow matching and diffusion models. arXiv preprint arXiv:2506.02070 (2025)

  20. [28]

    Hormann, K., Greiner,G.: Mips:An efficient globalparametrization method (2000)

  21. [29]

    ACM TOG33(4), 1–11 (2014)

    Huang, Q., Wang, F., Guibas, L.: Functional map networks for analyzing and exploring large shape collections. ACM TOG33(4), 1–11 (2014)

  22. [30]

    In: Computer Graphics Forum

    Huang, R., Ren, J., Wonka, P., Ovsjanikov, M.: Consistent zoomout: Efficient spectral map synchronization. In: Computer Graphics Forum. vol. 39, pp. 265–

  23. [31]

    Wiley Online Library (2020)

  24. [32]

    In: CVPR

    Jiang, H., Karpur, A., Cao, B., Huang, Q., Araujo, A.: Omniglue: Generalizable feature matching with foundation model guidance. In: CVPR. pp. 19865–19875 (2024)

  25. [33]

    ACM TOG30(4), 1–12 (2011)

    Kim, V.G., Lipman, Y., Funkhouser, T.: Blended intrinsic maps. ACM TOG30(4), 1–12 (2011)

  26. [34]

    In: Eurographics Workshop on 3D Object Retrieval, EG 3DOR

    Lähner, Z., Rodolà, E., Bronstein, M.M., Cremers, D., Burghard, O., Cosmo, L., Dieckmann, A., Klein, R., Sahillioˇ glu, Y., et al.: Shrec’16: Matching of deformable shapes with topological noise. In: Eurographics Workshop on 3D Object Retrieval, EG 3DOR. pp. 55–60. Eurographic...

  27. [35]

    NeurIPS35, 29336–29349 (2022)

    Li, L., Donati, N., Ovsjanikov, M.: Learning multi-resolution functional maps with spectral attention for robust shape matching. NeurIPS35, 29336–29349 (2022)

  28. [36]

    ACM TOG36(6), 194–1 (2017)

    Li, T., Bolkart, T., Black, M.J., Li, H., Romero, J.: Learning a model of facial shape and expression from 4d scans. ACM TOG36(6), 194–1 (2017)

  29. [37]

    In: CVPR

    Liang, F., Wu, B., Dai, X., Li, K., Zhao, Y., Zhang, H., Zhang, P., Vajda, P., Marculescu, D.: Open-vocabulary semantic segmentation with mask-adapted clip. In: CVPR. pp. 7061–7070 (2023)

  30. [38]

    arXiv preprint arXiv:2210.02747 (2022)

    Lipman, Y., Chen, R.T., Ben-Hamu, H., Nickel, M., Le, M.: Flow matching for generative modeling. arXiv preprint arXiv:2210.02747 (2022)

  31. [39]

    arXiv preprint arXiv:2412.06264 (2024)

    Lipman, Y., Havasi, M., Holderrieth, P., Shaul, N., Le, M., Karrer, B., Chen, R.T., Lopez-Paz, D., Ben-Hamu, H., Gat, I.: Flow matching guide and code. arXiv preprint arXiv:2412.06264 (2024)

  32. [40]

    In: ICCV

    Litany, O., Remez, T., Rodola, E., Bronstein, A., Bronstein, M.: Deep functional maps: Structured prediction for dense shape correspondence. In: ICCV. pp. 5659– 5667 (2017)

  33. [41]

    In: Computer Graphics Forum

    Litany, O., Rodolà, E., Bronstein, A.M., Bronstein, M.M.: Fully spectral partial shape matching. In: Computer Graphics Forum. vol. 36, pp. 247–258. Wiley Online Library (2017) SGMatch: Semantic-Guided Non-Rigid Shape Matching 17

  34. [42]

    In: Seminal Graphics Papers: Pushing the Boundaries, Volume 2, pp

    Loper, M., Mahmood, N., Romero, J., Pons-Moll, G., Black, M.J.: Smpl: A skinned multi-person linear model. In: Seminal Graphics Papers: Pushing the Boundaries, Volume 2, pp. 851–866 (2023)

  35. [43]

    IEEE TVCG (2025)

    Luo, F., Li, Q., Hu, L., Wang, H., Xu, H., Liu, X., Liu, S., Chen, H.: Deep frequency awareness functional maps for robust shape matching. IEEE TVCG (2025)

  36. [44]

    Magnet, R., Ren, J., Sorkine-Hornung, O., Ovsjanikov, M.: Smooth non-rigid shape matching via effective dirichlet energy optimization. In: 3DV. pp. 495–504. IEEE (2022)

  37. [45]

    In: Eurographics Workshop on 3D Object Retrieval

    Melzi, S., Marin, R., Rodolà, E., Castellani, U., Ren, J., Poulenard, A., Ovsjanikov, P., et al.: Shrec’19: matching humans with different connectivity. In: Eurographics Workshop on 3D Object Retrieval. pp. 1–8. The Eurographics Association (2019)

  38. [46]

    arXiv preprint arXiv:1904.07865 (2019)

    Melzi, S., Ren, J., Rodola, E., Sharma, A., Wonka, P., Ovsjanikov, M.: Zoomout: Spectral upsampling for efficient shape correspondence. arXiv preprint arXiv:1904.07865 (2019)

  39. [47]

    In: Computer Graphics Forum

    Morreale, L., Aigerman, N., Kim, V.G., Mitra, N.J.: Neural semantic surface maps. In: Computer Graphics Forum. vol. 43, p. e15005. Wiley Online Library (2024)

  40. [48]

    In: Computer Graphics Forum

    Nogneng, D., Ovsjanikov, M.: Informative descriptor preservation via commutativ- ity for shape matching. In: Computer Graphics Forum. vol. 36, pp. 259–267. Wiley Online Library (2017)

  41. [49]

    arXiv preprint arXiv:2304.07193 (2023)

    Oquab, M., Darcet, T., Moutakanni, T., Vo, H., Szafraniec, M., Khalidov, V., Fernandez, P., Haziza, D., Massa, F., El-Nouby, A., et al.: Dinov2: Learning robust visual features without supervision. arXiv preprint arXiv:2304.07193 (2023)

  42. [50]

    ACM TOG31(4), 1–11 (2012)

    Ovsjanikov, M., Ben-Chen, M., Solomon, J., Butscher, A., Guibas, L.: Functional maps: a flexible representation of maps between shapes. ACM TOG31(4), 1–11 (2012)

  43. [51]

    In: AAAI

    Perez, E., Strub, F., De Vries, H., Dumoulin, V., Courville, A.: Film: Visual rea- soning with a general conditioning layer. In: AAAI. vol. 32 (2018)

  44. [52]

    In: ICCV

    Pierson, E., Li, L., Dai, A., Ovsjanikov, M.: Diffumatch: Category-agnostic spectral diffusion priors for robust non-rigid shape matching. In: ICCV. pp. 5745–5756 (2025)

  45. [53]

    Experimental Mathematics2(1), 15–36 (1993)

    Pinkall,U.,Polthier,K.:Computingdiscreteminimalsurfacesandtheirconjugates. Experimental Mathematics2(1), 15–36 (1993)

  46. [54]

    In: ICML

    Radford, A., Kim, J.W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., et al.: Learning transferable visual models from natural language supervision. In: ICML. pp. 8748–8763. PmLR (2021)

  47. [55]

    In: Computer Graphics Forum

    Ren, J., Melzi, S., Wonka, P., Ovsjanikov, M.: Discrete optimization for shape matching. In: Computer Graphics Forum. vol. 40, pp. 81–96. Wiley Online Library (2021)

  48. [56]

    In: Computer Graphics Forum

    Ren, J., Panine, M., Wonka, P., Ovsjanikov, M.: Structured regularization of func- tional map computations. In: Computer Graphics Forum. vol. 38, pp. 39–53. Wiley Online Library (2019)

  49. [57]

    ACM TOG37(6), 1–16 (2018)

    Ren, J., Poulenard, A., Wonka, P., Ovsjanikov, M.: Continuous and orientation- preserving correspondences via functional maps. ACM TOG37(6), 1–16 (2018)

  50. [58]

    In: Computer Graphics Forum

    Rodolà, E., Cosmo, L., Bronstein, M.M., Torsello, A., Cremers, D.: Partial func- tional correspondence. In: Computer Graphics Forum. vol. 36, pp. 222–236. Wiley Online Library (2017)

  51. [59]

    The Visual Computer 36(8), 1705–1721 (2020)

    Sahillioğlu, Y.: Recent advances in shape correspondence. The Visual Computer 36(8), 1705–1721 (2020)

  52. [60]

    ACM TOG41(3), 1–16 (2022) 18 Tianwei Ye et al

    Sharp, N., Attaiki, S., Crane, K., Ovsjanikov, M.: Diffusionnet: Discretization ag- nostic learning on surfaces. ACM TOG41(3), 1–16 (2022) 18 Tianwei Ye et al

  53. [61]

    arXiv preprint arXiv:2508.10104 (2025)

    Siméoni, O., Vo, H.V., Seitzer, M., Baldassarre, F., Oquab, M., Jose, C., Khali- dov, V., Szafraniec, M., Yi, S., Ramamonjisoa, M., et al.: Dinov3. arXiv preprint arXiv:2508.10104 (2025)

  54. [62]

    NeurIPS34, 3108–3120 (2021)

    Song, C., Wei, J., Li, R., Liu, F., Lin, G.: 3d pose transfer with correspondence learning and mesh refinement. NeurIPS34, 3108–3120 (2021)

  55. [63]

    IEEE TPAMI45(8), 10488–10499 (2023)

    Song, C., Wei, J., Li, R., Liu, F., Lin, G.: Unsupervised 3d pose transfer with cross consistency and dual reconstruction. IEEE TPAMI45(8), 10488–10499 (2023)

  56. [64]

    In: CVPR

    Sun, M., Mao, S., Jiang, P., Ovsjanikov, M., Huang, R.: Spatially and spectrally consistent deep functional maps. In: CVPR. pp. 14497–14507 (2023)

  57. [65]

    NeurIPS34, 5731–5744 (2021)

    Trappolini, G., Cosmo, L., Moschella, L., Marin, R., Melzi, S., Rodolà, E.: Shape registration in the time of transformers. NeurIPS34, 5731–5744 (2021)

  58. [66]

    In: Computer Graphics Forum

    Van Kaick, O., Zhang, H., Hamarneh, G., Cohen-Or, D.: A survey on shape cor- respondence. In: Computer Graphics Forum. vol. 30, pp. 1681–1707. Wiley Online Library (2011)

  59. [67]

    In: AAAI

    Xia, Y., Lu, Y., Gao, Y., Ma, J.: Locality preserving refinement for shape matching with functional maps. In: AAAI. vol. 38, pp. 6207–6215 (2024)

  60. [68]

    In: AAAI

    Xia, Y., Ye, T., Zhou, H., Wang, Z., Ma, J.: Multi-shape matching with cycle consistency basis via functional maps. In: AAAI. vol. 39, pp. 8575–8583 (2025)

  61. [69]

    In: CVPR

    Xie, Y., Ehm, V., Roetzer, P., El Amrani, N., Gao, M., Bernard, F., Cremers, D.: Echomatch: Partial-to-partial shape matching via correspondence reflection. In: CVPR. pp. 11665–11675 (2025)

  62. [70]

    arXiv preprint arXiv:2509.16017 (2025)

    Yang, M., Fan, F., Li, Z., Deng, S., Ma, Y., Ma, J.: Distillmatch: Leveraging knowl- edge distillation from vision foundation model for multimodal image matching. arXiv preprint arXiv:2509.16017 (2025)

  63. [71]

    arXiv preprint arXiv:2509.01204 (2025)

    Ye, T., Ma, Y., Mei, X.: Dcmatch: Unsupervised multi-shape matching with dual- level consistency. arXiv preprint arXiv:2509.01204 (2025)

  64. [72]

    arXiv preprint arXiv:2511.15967 (2025)

    Yuan, M., Zhang, Y., Zhang, W., Ma, L., Gao, Y., Ying, J., Xin, Y.: Infoclip: Bridg- ing vision-language pretraining and open-vocabulary semantic segmentation via information-theoretic alignment transfer. arXiv preprint arXiv:2511.15967 (2025)

  65. [73]

    In: CVPR

    Zhang, Y., Zhao, X.: Mesa: Matching everything by segmenting anything. In: CVPR. pp. 20217–20226 (2024)

  66. [74]

    In: CVPR

    Zhuravlev, A., Lähner, Z., Golyanik, V.: Denoising functional maps: Diffusion mod- els for shape correspondence. In: CVPR. pp. 26899–26909 (2025)

  67. [75]

    In: CVPR

    Zuffi, S., Kanazawa, A., Jacobs, D.W., Black, M.J.: 3d menagerie: Modeling the 3d shape and pose of animals. In: CVPR. pp. 6365–6373 (2017) SGMatch: Semantic-Guided Non-Rigid Shape Matching 19 A Implementation Details A.1 Experimental Settings All learning-based methods are im...

Pith tools

Reviewed July 14, 2026 · model on record in the stance chip above.