Pith. sign in

REVIEW 4 major objections 6 minor 60 references

INRet: A General Framework for Accurate Retrieval of INRs for Shapes

T0 review · 4 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read INRet finds similar 3D shapes directly from neural-field encodings.

desk verdict First real retrieval method for grid-based INRs with clean cross-implicit ablations, but the headline accuracy comparison is confounded by architecture and needs a like-for-like rerun. read the letter →

arxiv 2501.15722 v1 pith:X6CVBWEC submitted 2025-01-27 cs.LG

Yushi Guan , Daniel Kwan , Ruofan Liang , Selvakumar Panneer , Nilesh Jain , Nilesh Ahuja , Nandita Vijaykumar This is my paper · ORCID
classification cs.LG
keywords implicitneuralrepresentation3Dshaperetrievalfeaturegridsoctreetriplanehashgridembeddingsigneddistancefunction
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

INRet sets out to answer a practical question: once 3D shapes are stored as implicit neural representations (INRs), how do you retrieve similar shapes from a data store? The paper claims that similarity can be computed directly from the INR itself, using both its MLP weights and its learned feature-grid parameters, instead of converting the INR to a point cloud or multi-view images. The method targets the feature-grid architectures used in practice today, namely octree, triplane, and hash grids, and it is designed to handle the three common implicit functions: signed distance, unsigned distance, and occupancy. If correct, this means an INR data store can be searched by another INR, even across different architectures and different implicit functions, with higher accuracy than conversion-based retrieval and with much lower latency. On ShapeNet10 and Pix3D, grid-based SDF INRs reach 82–84% and 74–78% mAP@1, respectively, roughly 12–15% above the inr2vec, PointNeXt, and View-GCN baselines for iNGP INRs.

What carries the argument

The load-bearing object is the INR Embedding Encoder, composed of an MLP encoder for the INR's network weights and a Conv3D encoder for its feature grid. For the grid, the paper samples $(2N)^3$ feature vectors at a fixed resolution, summing features across octree or triplane levels with zero-filling for missing voxels, or reading hash-grid features through the original hash function, then passes the sampled volume through 3D convolutions. The second mechanism is the unified latent space: an explicit L2 loss and a Unified Shape Decoder force embeddings from different implicit functions of the same shape to converge to one representation, so cosine similarity in that space can be compared across SDF, UDF, and occupancy INRs.

What would settle it

Take a pair of different shapes whose feature grids agree at the chosen $(2N)^3$ sampling locations, for instance two objects that differ only in a thin region between grid points, and train INRs for them; if INRet ranks this pair as near-duplicates while a higher-resolution sampling or a point-cloud conversion separates them cleanly, the fixed-resolution sampling is the failure point. A quantitative version is to measure mAP@1 as $N$ grows: if accuracy is still improving at the largest practical $N$, the fixed resolution is a ceiling on the method rather than a neutral implementation detail.

Watch

Extended reading notes

Core claim

The central claim is that a shape INR can be converted into a fixed-length embedding that preserves enough information for accurate retrieval by encoding the INR's MLP weights with an MLP encoder and encoding a fixed-resolution sample of its feature grid with a Conv3D encoder, then concatenating the two. To make embeddings comparable across implicit functions, the paper trains separate encoders per implicit function and regularizes them in two ways: an explicit L2 loss pulls embeddings of the same shape together across SDF, UDF, and occupancy representations, and a single Unified Shape Decoder supervises all three encoders to reconstruct one common implicit function. The paper claims this enables accurate retrieval across grid architectures and implicit functions, and reports that it outperforms the prior INR-retrieval method inr2vec, which only supports MLP-only INRs with matching architectures, as well as point-cloud and multi-view-image retrieval on converted INRs.

Load-bearing premise

The load-bearing premise is that a fixed-resolution $(2N)^3$ sampling of the feature grid, with octree levels summed, missing voxels zero-filled, and hash or triplane grids read through their native lookups, keeps enough geometric information to tell different shapes apart; if that sampling loses the discriminating detail or makes different architectures look alike, the Conv3D encoder cannot build comparable embeddings.

Editorial extensions

If this is right

  • An INR data store can be searched directly with an INR query, avoiding point-cloud or multi-view conversion and cutting retrieval latency by one to two orders of magnitude in the no-conversion case.
  • Retrieval works across feature-grid architectures: a query MLP-only INR can be distilled into a supported grid architecture and still retrieve NGLOD, EG3D, or iNGP INRs with accuracy comparable to or better than inr2vec.
  • With both regularizations, querying a UDF, SDF, or occupancy INR retrieves INRs of all three types at roughly the same accuracy as same-type retrieval, about 82% average for iNGP on ShapeNet10, whereas inr2vec drops to near 10% for cross-type retrieval.
  • Because grid-based INRs reconstruct shapes more faithfully than MLP-only INRs, retrieval built on them inherits that fidelity; the paper shows even under-trained iNGP INRs outperform MLP-only INRs in retrieval accuracy.
  • New grid-based INR architectures can be supported either by training an analogous sampling-based encoder or by distilling the new architecture into a supported one.

Reading between the lines

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

  • A consequence the authors leave implicit is that the fixed-resolution grid sampling defines, in effect, a canonical tensor view of a broad family of grid INRs, so the same Conv3D encoder could be reused for any future grid whose features can be sampled the same way.
  • The paper's appendix shows that cosine-similarity embeddings score only 21–28% on its own Category-Chamfer metric, which rewards retrieving the exact instance with smallest Chamfer distance rather than just the same category; this suggests the central claim is best read as category-level semantic retrieval, with instance-level geometric matching left as an open problem.
  • The cross-architecture results imply that retrieval accuracy depends on respecting each architecture's native feature-combination rule: summing iNGP hash-grid features instead of concatenating them drops mAP from 84.2 to 30.4, so future encoders should preserve each grid's original aggregation rather than imposing one rule.
  • An untested extension is to vary the sampling resolution $N$ and measure mAP: if accuracy keeps rising with $N$, the fixed resolution is the bottleneck, and if it saturates early, the encoder or the INR representation itself is the limiting factor.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes INRet, a framework for retrieving 3D shapes that are represented as implicit neural representations (INRs). The method encodes both the MLP weights and the feature grid of an INR into an embedding via an MLP encoder and a Conv3D encoder, respectively, and uses cosine similarity between embeddings for retrieval. To support different implicit functions (SDF, UDF, occupancy), the authors train separate encoders and add two regularizations: an L2 loss that pulls together embeddings of the same shape across implicit functions, and a unified shape decoder that predicts a single implicit function from all embeddings. Experiments on ShapeNet10 and Pix3D compare INRet against inr2vec, PointNeXt, and View-GCN, including cross-architecture retrieval via INR distillation and extensive ablations over design choices. The paper also analyzes the relationship between reconstruction quality and retrieval accuracy.

Significance. The core contribution is timely and practically relevant: it is, to my knowledge, the first INR retrieval method that directly handles feature-grid-based INRs (octree, triplane, hash grid) and enables retrieval across different implicit functions, avoiding explicit conversion to point clouds or images. The paper is unusually thorough in its ablations (e.g., unified decoder choice, L2 weighting, norm choice, feature combination sensitivity, reconstruction quality analysis), and it honestly reports failure cases and the limited performance on the proposed Category-Chamfer metric. If the headline claims survive a fair architecture-matched comparison, INRet would be a solid benchmark for INR retrieval.

major comments (4)
  1. [§5.2, Tab. 1] The headline comparison confounds retrieval method with INR architecture. INRet is evaluated on NGLOD/EG3D/iNGP grid-based INRs, while inr2vec is evaluated on MLP-only SIREN INRs. The reconstruction-quality measurements in Tab. 15 show that the grid-based INRs are substantially more accurate (C.D. 0.012–0.022 vs 0.034–0.039 for MLP), and Tab. 16 shows that when iNGP is undertrained to 2 epochs (C.D. 0.0371, close to the MLP's 0.0354), its mAP drops from 84.2 to 78.8, still above but much closer to inr2vec's 73.4. This suggests the reported 12.0% accuracy gain may largely reflect representation quality rather than a superior retrieval mechanism. Please add a like-for-like comparison: e.g., INRet on MLP-only INRs (as in Tab. 10) and inr2vec on grid-based INRs if it can be adapted, or at least an ablation that controls for reconstruction quality (e.g., matching C.D. across compared architectures).
  2. [§3.2, Tab. 14] The generality claim for arbitrary grid architectures is undermined by the sensitivity of the Conv3D encoder to the native feature-combination convention. Switching from summation to concatenation for NGLOD drops mAP from 82.6 to 67.8 (a 14.8-point drop), and switching from concatenation to summation for iNGP drops it from 84.2 to 30.4 (a 53.8-point drop). This indicates that the encoder learns a representation that is tightly coupled to the specific grid's feature-combination rule, rather than a representation-agnostic geometric signature. The claim in §3.2 that 'a similar feature grid sampling approach can be used to also train an encoder for any new grid-based architecture' should be qualified, and the paper should discuss whether the encoder needs retraining for each new combination convention.
  3. [Abstract and §5.2] The headline improvement numbers are internally inconsistent and do not match Tab. 1. The abstract states a 10.1% average improvement and the conclusion repeats 'more than 10%'; §5.2 reports 12.0% higher accuracy than inr2vec for iNGP, which itself differs from the 10.8-point gap in Tab. 1 (84.2 vs 73.4 on ShapeNet10). Please state the exact numbers, specify whether the figures are percentage points or relative percentages, and recompute the average from the tables.
  4. [All tables] No error bars or statistical significance tests are reported, and no code is provided. Since ShapeNet10 has only 50 test shapes per category and retrieval results are reported at a single seed, the 5–10 point gaps against baselines could be within run-to-run variance. Please report results over multiple seeds (at least for the main tables) or release code to make the numerical claims reproducible.
minor comments (6)
  1. [§5.2] The phrase '12.0%, 15.4%, and 12.6% higher accuracy' should be clarified as percentage-point differences rather than relative improvements, since the compared mAP values are already percentages.
  2. [Eq. (4)] The loss in Eq. (4) is written with an absolute value where an L1 norm is intended; please define the norm explicitly and distinguish the L1 loss for the unified decoder from the L2 regularization term in the same equation.
  3. [Tabs. 3 and 7] The legend explaining the multiple numbers per cell is given only in the table text, not in the caption; please include it in the caption so the reader can interpret the entries without scanning the body text.
  4. [§5.3] The paper reports that INR distillation introduces approximately 30 seconds of overhead; please state whether this is a one-time cost per query or per INR, and clarify how it should be weighed against the retrieval latencies in Tab. 1.
  5. [§3.2] The sampling resolution N for the Conv3D encoder is a free parameter, but no sensitivity analysis with respect to N is reported; a brief ablation or a justification for the chosen value would strengthen the presentation.
  6. [Fig. 1 caption] The sentence 'The encoders take MLP weights and feature grid parameters as inputs to generate the INR Embedding' is slightly misleading because the decoder is used only during training; please rephrase to make clear that the decoder is not used at inference.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the method's training objectives explicitly produce the unified embedding space, and all headline retrieval metrics are reported on held-out test shapes.

full rationale

The paper's derivation chain is self-contained. INRet trains its MLP and Conv3D encoders with a reconstruction loss (Eq. 3) and an explicit L2 alignment loss between embeddings of the same shape encoded with different implicit functions (Eq. 4), with the shape decoder used only during training. Retrieval is evaluated on held-out test shapes from ShapeNet10 and Pix3D, and the cross-implicit-function experiments exclude the same underlying shape from the retrieval candidates (App. 8.2), so the reported mAP reflects generalization to unseen shapes rather than being forced by the training loss. The architecture-specific feature combination shown in App. 8.7 and the reconstruction-quality dependence shown in Sec. 9.2 are limitations on the generality claim, but they are not circularities: no fitted parameter is renamed as a prediction, no load-bearing step is equivalent to its inputs by definition, and the paper does not rely on self-citations or imported uniqueness theorems. The baseline comparison confounds INR architecture with retrieval method, which is a validity concern for the claimed accuracy margin, but that is a benchmarking issue, not a circular derivation.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central method relies on learning embeddings from INR weights and grids. It introduces no new physical entities, but has several hand-chosen hyperparameters (lambda, sampling resolution N) and relies on domain assumptions about the relationship between implicit functions and about INR fidelity. The most important unstated parameter is the grid sampling resolution N.

free parameters (3)
  • lambda (L2 regularization weight) = 1
    In Eq. 4, the balance between reconstruction loss and embedding alignment loss is set to 1, described as 'works well in practice'. Robustness is tested in App. 8.5, but the default is a hand-chosen hyperparameter.
  • Feature grid sampling resolution N = unspecified
    The sampling grid is defined as (2N)^3 in Sec. 3.2, but the actual value of N used in experiments is not reported, which is a hidden free parameter affecting the Conv3D encoder input.
  • Hierarchical sampling threshold = 3x smallest Chamfer distance
    In App. 11.3, the threshold for re-evaluating candidates at 4096 points is set to 3 times the smallest Chamfer distance at 128 points. This is an ad hoc parameter for the additional retrieval scheme, not the core method.
assumptions (4)
  • domain assumption The implicit functions UDF, SDF, and Occ of the same shape are related by simple operations (Eq. 8-11), so a shared decoder can reconstruct all three from aligned embeddings.
    Used in Sec. 3.3 and 7.5 to justify training separate encoders with a Unified Shape Decoder and L2 alignment; if this relation fails for imperfectly learned INRs, the alignment objective could distort embeddings.
  • domain assumption Learned INR weights and feature grids encode enough geometric information that a fixed grid sampling and weight flattening can produce discriminative embeddings.
    Core premise of the encoder design in Sec. 3.2; if the INR itself does not faithfully encode the shape, retrieval accuracy will degrade, as the paper's own Tab. 16 shows for undertrained INRs.
  • domain assumption Cosine similarity between the learned embeddings is a valid proxy for shape similarity and category membership.
    Used as the retrieval criterion in Sec. 5; the paper evaluates this indirectly via category-based mAP, not by direct shape-geometry judgment.
  • domain assumption Category labels from ShapeNet10 and Pix3D are an appropriate ground truth for shape similarity.
    All main retrieval metrics are category-based (Sec. 5.1); the paper partially addresses this with the Category-Chamfer metric in App. 11, where performance is markedly lower.

how reviews work

0 comments
Cite this review

Pith. "Pith review of INRet: A General Framework for Accurate Retrieval of INRs for Shapes." pith.science (2026). https://pith.science/paper/X6CVBWEC

@misc{pith2026250115722,
  author       = {Pith},
  title        = {Pith review of: INRet: A General Framework for Accurate Retrieval of INRs for Shapes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/X6CVBWEC}},
  note         = {Machine review of arXiv:2501.15722}
}
read the original abstract

Implicit neural representations (INRs) have become an important method for encoding various data types, such as 3D objects or scenes, images, and videos. They have proven to be particularly effective at representing 3D content, e.g., 3D scene reconstruction from 2D images, novel 3D content creation, as well as the representation, interpolation, and completion of 3D shapes. With the widespread generation of 3D data in an INR format, there is a need to support effective organization and retrieval of INRs saved in a data store. A key aspect of retrieval and clustering of INRs in a data store is the formulation of similarity between INRs that would, for example, enable retrieval of similar INRs using a query INR. In this work, we propose INRet, a method for determining similarity between INRs that represent shapes, thus enabling accurate retrieval of similar shape INRs from an INR data store. INRet flexibly supports different INR architectures such as INRs with octree grids, triplanes, and hash grids, as well as different implicit functions including signed/unsigned distance function and occupancy field. We demonstrate that our method is more general and accurate than the existing INR retrieval method, which only supports simple MLP INRs and requires the same architecture between the query and stored INRs. Furthermore, compared to converting INRs to other representations (e.g., point clouds or multi-view images) for 3D shape retrieval, INRet achieves higher accuracy while avoiding the conversion overhead.

Figures

Figures reproduced from arXiv: 2501.15722 by the authors.

Figure 1
Figure 1. Encoders trained to generate embeddings for grid-based INRs: INRs with hash-table/octree/triplane based feature grids are used to generate embeddings for similarity calculations using encoders (m and c). The encoders take MLP weights and feature grid parameters as inputs to generate the INR Embedding. During training, the encoders (m, c) and the decoder (fϕ) are jointly trained: the encoders to produce the INR Embed… view at source ↗
Figure 2
Figure 2. INR Embed. Creation for INRs with Different Implicit Functions. (a) For each shape, we train INRs with different implicit functions. (b) We train different encoders for INRs with different implicit functions. The differences between embeddings created by the encoders are minimized by L2 loss. (c) We feed the embeddings into a Unified Shape Decoder to recreate the UDF of the original shape. cal tracing from randomly … view at source ↗
Figure 4
Figure 4. INR Embedding tSNE Plot 6. Conclusion In this work, we presented a new framework for determin￾ing similarity between INRs that can be used for accurate retrieval of INRs from a data store. We proposed a new encoding method for INRs with feature grids including the octree and hash table based grids. By using L2 loss and a common decoder as regularizations, INRet also enables the retrieval of INRs across different imp… view at source ↗
Figures from the paper (4 more)
Figure 5
Figure 5. Figure 5: INR Embedding Creation for INRs with Different Architectures (a) Single Layer Ball/Cylinder UDF (d) Double Layer Cylinder UDF (b) Single Layer Ball SDF (c) Single Layer Ball Occ [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]
Figure 6
Figure 6. Figure 6: Implicit Function Visual Representation for Cross Section of Different Shapes our decoder MLP fϕ has the same architecture as in [13]. Only the encoders are necessary for generating the INR em￾beddings. The decoder is used solely during the training of the encoders, an…
Figure 7
Figure 7. Figure 7: Chair Retrieval Incorrect Classes 9. the Impact of Reconstruction Quality on Re￾trieval Accuracy 9.1. Reconstruction Quality In this section, we provide additional details on the qual￾ity of reconstruction of the trained INRs with respect to the original mesh. For UDF …
Figure 8
Figure 8. Figure 8: Hierarchical Sampling Retrieval Method Method Ours inr2vec Input Type iNGP iNGP @ 2 Epoch MLP INR mAP @ 1 84.2 78.8 73.4 C.D. 0.0168 0.0371 0.0354 [PITH_FULL_IMAGE:figures/full_fig_p018_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

60 extracted references · 48 canonical work pages

  1. [1]

    Sal: Sign agnos- tic learning of shapes from raw data

    Matan Atzmon and Yaron Lipman. Sal: Sign agnos- tic learning of shapes from raw data. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2565–2574, 2020. 3

    work page 2020

  2. [2]

    Sal++: Sign agnostic learning with derivatives

    Matan Atzmon and Yaron Lipman. Sal++: Sign agnostic learning with derivatives. ArXiv, abs/2006.05400, 2020. 3

    arXiv 2006

  3. [3]

    Kt-net: knowledge transfer for unpaired 3d shape completion

    Zhen Cao, Wenxiao Zhang, Xin Wen, Zhen Dong, Yu-Shen Liu, Xiongwu Xiao, and Bisheng Yang. Kt-net: knowledge transfer for unpaired 3d shape completion. InProceedings of the AAAI Conference on Artificial Intelligence , pages 286– 294, 2023. 1

    work page 2023

  4. [4]

    Lin, Matthew Chan, Koki Nagano, Boxiao Pan, Shalini De Mello, Orazio Gallo, Leonidas J

    Eric Chan, Connor Z. Lin, Matthew Chan, Koki Nagano, Boxiao Pan, Shalini De Mello, Orazio Gallo, Leonidas J. Guibas, Jonathan Tremblay, S. Khamis, Tero Karras, and Gordon Wetzstein. Efficient geometry-aware 3d generative adversarial networks. 2022 IEEE/CVF Conference on Com- puter Vision and Pattern Recognition (CVPR), pages 16102– 16112, 2021. 2, 4, 6

    work page 2022

  5. [5]

    Chang, Thomas A

    Angel X. Chang, Thomas A. Funkhouser, Leonidas J. Guibas, Pat Hanrahan, Qi-Xing Huang, Zimo Li, Silvio Savarese, Manolis Savva, Shuran Song, Hao Su, Jianxiong Xiao, L. Yi, and Fisher Yu. Shapenet: An information-rich 3d model repository. ArXiv, abs/1512.03012, 2015. 6

    arXiv 2015

  6. [6]

    Tensorf: Tensorial radiance fields

    Anpei Chen, Zexiang Xu, Andreas Geiger, Jingyi Yu, and Hao Su. Tensorf: Tensorial radiance fields. ArXiv, abs/2203.09517, 2022. 4

    arXiv 2022

  7. [7]

    Learning con- tinuous image representation with local implicit image func- tion

    Yinbo Chen, Sifei Liu, and Xiaolong Wang. Learning con- tinuous image representation with local implicit image func- tion. 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 8624–8634, 2020. 1

    work page 2021

  8. [8]

    Learning implicit fields for generative shape modeling

    Zhiqin Chen and Hao Zhang. Learning implicit fields for generative shape modeling. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 5939–5948, 2019. 1, 3

    work page 2019

Show all 60 references
  1. [9]

    Videoinr: Learning video implicit neural representa- tion for continuous space-time super-resolution

    Zeyuan Chen, Yinbo Chen, Jingwen Liu, Xingqian Xu, Vidit Goel, Zhangyang Wang, Humphrey Shi, and Xiaolong Wang. Videoinr: Learning video implicit neural representa- tion for continuous space-time super-resolution. In Proceed- ings of the IEEE/CVF Conference on Computer Vision ...

    2022

  2. [10]

    Tulyakov, Alexander G

    Yen-Chi Cheng, Hsin-Ying Lee, S. Tulyakov, Alexander G. Schwing, and Liangyan Gui. Sdfusion: Multimodal 3d shape completion, reconstruction, and generation. 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 4456–4465, 2022. 2

    2023

  3. [11]

    Neu- ral unsigned distance fields for implicit function learning

    Julian Chibane, Aymen Mir, and Gerard Pons-Moll. Neu- ral unsigned distance fields for implicit function learning. ArXiv, abs/2010.13938, 2020. 2, 3, 4, 6, 1, 7

    2010 arXiv

  4. [12]

    Davies, Derek Nowrouzezahrai, and Alec Jacobson

    T. Davies, Derek Nowrouzezahrai, and Alec Jacobson. Over- fit neural networks as a compact shape representation.ArXiv, abs/2009.09808, 2020. 4

    2009 arXiv

  5. [13]

    Deep learning on implicit neural representations of shapes

    Luca De Luigi, Adriano Cardace, Riccardo Spezialetti, Pier- luigi Zama Ramirez, Samuele Salti, and Luigi di Stefano. Deep learning on implicit neural representations of shapes. In International Conference on Learning Representations ,

  6. [14]

    Coin: Compression with implicit neural representations

    Emilien Dupont, Adam Goli ´nski, Milad Alizadeh, Yee Whye Teh, and Arnaud Doucet. Coin: Compression with implicit neural representations. arXiv preprint arXiv:2103.03123 ,

    arXiv

  7. [15]

    Ue4-nerf: Neural radiance field for real-time rendering of large-scale scene

    Jiaming Gu, Minchao Jiang, Hongsheng Li, Xiaoyuan Lu, Guangming Zhu, Syed Afaq Ali Shah, Liang Zhang, and Mohammed Bennamoun. Ue4-nerf: Neural radiance field for real-time rendering of large-scale scene. Advances in Neural Information Processing Systems, 36, 2024. 1

    2024

  8. [16]

    Meshcnn: a network with an edge

    Rana Hanocka, Amir Hertz, Noa Fish, Raja Giryes, Shachar Fleishman, and Daniel Cohen-Or. Meshcnn: a network with an edge. ACM Transactions on Graphics (TOG), 38:1 – 12,

  9. [17]

    John C. Hart. Sphere tracing: a geometric method for the antialiased ray tracing of implicit surfaces. The Visual Com- puter, 12:527–545, 1996. 5

    1996

  10. [18]

    An accurate method for voxelizing polygon meshes

    Jian Huang, Roni Yagel, Vassily Filippov, and Yair Kurzion. An accurate method for voxelizing polygon meshes. IEEE Symposium on Volume Visualization (Cat. No.989EX300) , pages 119–126, 1998. 2

    1998

  11. [19]

    Batch normalization: Accelerating deep network training by reducing internal co- variate shift

    Sergey Ioffe and Christian Szegedy. Batch normalization: Accelerating deep network training by reducing internal co- variate shift. In International Conference on Machine Learn- ing, 2015. 1

    2015

  12. [20]

    Neuralvdb: High-resolution sparse volume representation using hierar- chical neural networks

    Doyub Kim, Minjae Lee, and Ken Museth. Neuralvdb: High-resolution sparse volume representation using hierar- chical neural networks. ArXiv, abs/2208.04448, 2022. 4

    2022 arXiv

  13. [21]

    Kingma and Jimmy Ba

    Diederik P. Kingma and Jimmy Ba. Adam: A method for stochastic optimization. CoRR, abs/1412.6980, 2014. 1, 6

    2014 arXiv

  14. [22]

    Meshwalker: Deep mesh under- standing by random walks

    Alon Lahav and Ayellet Tal. Meshwalker: Deep mesh under- standing by random walks. ACM Trans. Graph., 39:263:1– 263:13, 2020. 1, 4

    2020

  15. [23]

    Lorensen and Harvey E

    William E. Lorensen and Harvey E. Cline. Marching cubes: A high resolution 3d surface construction algorithm. Pro- ceedings of the 14th annual conference on Computer graph- ics and interactive techniques, 1987. 2, 6

    1987

  16. [24]

    Decoupled weight de- cay regularization

    Ilya Loshchilov and Frank Hutter. Decoupled weight de- cay regularization. In International Conference on Learning Representations, 2017. 1

    2017

  17. [25]

    Occupancy networks: Learning 3d reconstruction in function space

    Lars Mescheder, Michael Oechsle, Michael Niemeyer, Se- bastian Nowozin, and Andreas Geiger. Occupancy networks: Learning 3d reconstruction in function space. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 4460–4470, 2019. 2, 3, 4, 6

    2019

  18. [26]

    Srinivasan, Matthew Tancik, Jonathan T

    Ben Mildenhall, Pratul P. Srinivasan, Matthew Tancik, Jonathan T. Barron, Ravi Ramamoorthi, and Ren Ng. Nerf. Communications of the ACM, 65:99 – 106, 2020. 1

    2020

  19. [27]

    Mitchel, Vladimir G

    Thomas W. Mitchel, Vladimir G. Kim, and Michael M. Kazhdan. Field convolutions for surface cnns. 2021 IEEE/CVF International Conference on Computer Vision (ICCV), pages 9981–9991, 2021. 1, 4

    2021

  20. [28]

    Cheng, Maneesh Singh, and Shubham Tulsiani

    Paritosh Mittal, Y . Cheng, Maneesh Singh, and Shubham Tulsiani. Autosdf: Shape priors for 3d completion, recon- struction and generation. 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages 306–315, 2022. 2

    2022

  21. [29]

    Instant neural graphics primitives with a mul- tiresolution hash encoding

    Thomas M ¨uller, Alex Evans, Christoph Schied, and Alexan- der Keller. Instant neural graphics primitives with a mul- tiresolution hash encoding. ACM Transactions on Graphics (ToG), 41(4):1–15, 2022. 2, 4, 6

    2022

  22. [30]

    Deepsdf: Learning con- tinuous signed distance functions for shape representation

    Jeong Joon Park, Peter Florence, Julian Straub, Richard Newcombe, and Steven Lovegrove. Deepsdf: Learning con- tinuous signed distance functions for shape representation. In Proceedings of the IEEE/CVF conference on computer vi- sion and pattern recognition, pages 165–174, 201...

    2019

  23. [31]

    Mescheder, Marc Pollefeys, and Andreas Geiger

    Songyou Peng, Michael Niemeyer, Lars M. Mescheder, Marc Pollefeys, and Andreas Geiger. Convolutional occu- pancy networks. ArXiv, abs/2003.04618, 2020. 4

    2003 arXiv

  24. [32]

    Barron, and Ben Milden- hall

    Ben Poole, Ajay Jain, Jonathan T. Barron, and Ben Milden- hall. Dreamfusion: Text-to-3d using 2d diffusion. ArXiv, abs/2209.14988, 2022. 1

    2022 arXiv

  25. [33]

    Qi, Hao Su, Kaichun Mo, and Leonidas J

    C. Qi, Hao Su, Kaichun Mo, and Leonidas J. Guibas. Point- net: Deep learning on point sets for 3d classification and seg- mentation. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 77–85, 2016. 1, 4

    2017

  26. [34]

    C. Qi, L. Yi, Hao Su, and Leonidas J. Guibas. Pointnet++: Deep hierarchical feature learning on point sets in a metric space. In NIPS, 2017

    2017

  27. [35]

    Pointnext: Revisiting pointnet++ with improved training and scaling strategies

    Guocheng Qian, Yuchen Li, Houwen Peng, Jinjie Mai, Hasan Abed Al Kader Hammoud, Mohamed Elhoseiny, and Bernard Ghanem. Pointnext: Revisiting pointnet++ with improved training and scaling strategies. ArXiv, abs/2206.04670, 2022. 1, 4, 6, 2

    2022 arXiv

  28. [36]

    Deep learning on 3d neural fields, 2023

    Pierluigi Zama Ramirez, Luca De Luigi, Daniele Sirocchi, Adriano Cardace, Riccardo Spezialetti, Francesco Ballerini, Samuele Salti, and Luigi Di Stefano. Deep learning on 3d neural fields, 2023. 2

    2023

  29. [37]

    Shrec16 track: largescale 3d shape retrieval from shapenet core55

    Manolis Savva, Fisher Yu, Hao Su, M Aono, B Chen, D Cohen-Or, W Deng, Hang Su, Song Bai, Xiang Bai, et al. Shrec16 track: largescale 3d shape retrieval from shapenet core55. In Proceedings of the eurographics workshop on 3D object retrieval, 2016. 6, 4, 7

    2016

  30. [38]

    Retrievalfuse: Neural 3d scene reconstruction with a database

    Yawar Siddiqui, Justus Thies, Fangchang Ma, Qi Shan, Matthias Nießner, and Angela Dai. Retrievalfuse: Neural 3d scene reconstruction with a database. In Proceedings of the IEEE/CVF International Conference on Computer Vi- sion, pages 12568–12577, 2021. 1

    2021

  31. [39]

    Implicit neural representa- tions with periodic activation functions

    Vincent Sitzmann, Julien Martel, Alexander Bergman, David Lindell, and Gordon Wetzstein. Implicit neural representa- tions with periodic activation functions. Advances in Neural Information Processing Systems, 33:7462–7473, 2020. 2, 3, 1

    2020

  32. [40]

    Inras: Implicit neural representation for audio scenes

    Kun Su, Mingfei Chen, and Eli Shlizerman. Inras: Implicit neural representation for audio scenes. In Neural Informa- tion Processing Systems, 2022. 1

    2022

  33. [41]

    Tenen- baum, and William T

    Xingyuan Sun, Jiajun Wu, Xiuming Zhang, Zhoutong Zhang, Chengkai Zhang, Tianfan Xue, Joshua B. Tenen- baum, and William T. Freeman. Pix3d: Dataset and methods for single-image 3d shape modeling. 2018 IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 2974–...

    2018

  34. [42]

    Neural geometric level of detail: Real-time rendering with implicit 3d shapes

    Towaki Takikawa, Joey Litalien, Kangxue Yin, Karsten Kreis, Charles Loop, Derek Nowrouzezahrai, Alec Jacobson, Morgan McGuire, and Sanja Fidler. Neural geometric level of detail: Real-time rendering with implicit 3d shapes. In Proceedings of the IEEE/CVF Conference on Computer...

    2021

  35. [43]

    Variable bitrate neural fields

    Towaki Takikawa, Alex Evans, Jonathan Tremblay, Thomas M¨uller, Morgan McGuire, Alec Jacobson, and Sanja Fidler. Variable bitrate neural fields. InACM SIGGRAPH 2022 Con- ference Proceedings, pages 1–9, 2022. 4

    2022

  36. [44]

    Kaolin wisp: A pytorch library and en- gine for neural fields research

    Towaki Takikawa, Or Perel, Clement Fuji Tsang, Charles Loop, Joey Litalien, Jonathan Tremblay, Sanja Fidler, and Maria Shugrina. Kaolin wisp: A pytorch library and en- gine for neural fields research. https://github.com/ NVIDIAGameWorks/kaolin-wisp, 2022. 1

    2022

  37. [45]

    Nerfstudio: A modu- lar framework for neural radiance field development

    Matthew Tancik, Ethan Weber, Evonne Ng, Ruilong Li, Brent Yi, Justin Kerr, Terrance Wang, Alexander Kristof- fersen, Jake Austin, Kamyar Salahi, Abhik Ahuja, David McAllister, and Angjoo Kanazawa. Nerfstudio: A modu- lar framework for neural radiance field development. ACM SIG...

    2023

  38. [46]

    Peng-Shuai Wang, Yang Liu, Yu-Xiao Guo, Chun-Yu Sun, and Xin Tong. O-cnn. ACM Transactions on Graphics (TOG), 36:1 – 11, 2017. 1, 4

    2017

  39. [47]

    Adaptive o-cnn: A patch-based deep representation of 3d shapes

    Peng-Shuai Wang, Chun-Yu Sun, Yang Liu, and Xin Tong. Adaptive o-cnn: A patch-based deep representation of 3d shapes. arXiv: Computer Vision and Pattern Recognition ,

  40. [48]

    View-gcn: View- based graph convolutional network for 3d shape analysis

    Xin Wei, Ruixuan Yu, and Jian Sun. View-gcn: View- based graph convolutional network for 3d shape analysis. 2020 IEEE/CVF Conference on Computer Vision and Pat- tern Recognition (CVPR), pages 1847–1856, 2020. 4, 6, 2

    2020

  41. [49]

    Wolfe and Keld T

    Cameron R. Wolfe and Keld T. Lundgaard. E-stitchup: Data augmentation for pre-trained embeddings. arXiv: Learning,

  42. [50]

    Group normalization

    Yuxin Wu and Kaiming He. Group normalization. Inter- national Journal of Computer Vision, 128:742 – 755, 2018. 1

    2018

  43. [51]

    Pix2vox: Context-aware 3d reconstruction from single and multi-view images

    Haozhe Xie, Hongxun Yao, Xiaoshuai Sun, Shangchen Zhou, Shengping Zhang, and Xiaojun Tong. Pix2vox: Context-aware 3d reconstruction from single and multi-view images. 2019 IEEE/CVF International Conference on Com- puter Vision (ICCV), pages 2690–2698, 2019. 2

    2019

  44. [52]

    Neural fields in visual computing and beyond

    Yiheng Xie, Towaki Takikawa, Shunsuke Saito, Or Litany, Shiqin Yan, Numair Khan, Federico Tombari, James Tomp- kin, Vincent Sitzmann, and Srinath Sridhar. Neural fields in visual computing and beyond. InComputer Graphics Forum, pages 641–676. Wiley Online Library, 2022. 4

    2022

  45. [53]

    Road: Learning an implicit recursive octree auto-decoder to efficiently encode 3d shapes

    Sergey Zakharov, Rares Ambrus, Katherine Liu, and Adrien Gaidon. Road: Learning an implicit recursive octree auto-decoder to efficiently encode 3d shapes. ArXiv, abs/2212.06193, 2022. 4

    2022 arXiv

  46. [54]

    Implicit neural video com- pression

    Yunfan Zhang, Ties van Rozendaal, Johann Brehmer, Markus Nagel, and Taco Cohen. Implicit neural video com- pression. arXiv preprint arXiv:2112.11312, 2021. 1

    2021 arXiv

  47. [55]

    Learning consistency-aware unsigned dis- tance functions progressively from raw point clouds

    Junsheng Zhou, Baorui Ma, Yu-Shen Liu, Yi Fang, and Zhizhong Han. Learning consistency-aware unsigned dis- tance functions progressively from raw point clouds. ArXiv, abs/2210.02757, 2022. 2, 3 INRet: A General Framework for Accurate Retrieval of INRs for Shapes Supplementary Material

    2022 arXiv

  48. [56]

    INR Architecture and Training Detail INR Training Losses

    Architecture and Training Details 7.1. INR Architecture and Training Detail INR Training Losses. We apply different loss functions for different implicit functions. For the signed distance func- tion, we follow the method in [42]. Ls(fθ(x), ds(x)) = ∥fθ(x) − ds(x)∥2 (5) For th...

  49. [57]

    Additional Results In this section of the appendix, we provide additional re- sults and ablation studies for INRet. App. 8.1 and 8.2 pro- vides additional results for retrieval accuracy evaluation on ShapeNet10 and Pix3D. App. 8.3 demonstrates the effec- tiveness of INRet’s re...

  50. [58]

    Reconstruction Quality In this section, we provide additional details on the qual- ity of reconstruction of the trained INRs with respect to the original mesh

    the Impact of Reconstruction Quality on Re- trieval Accuracy 9.1. Reconstruction Quality In this section, we provide additional details on the qual- ity of reconstruction of the trained INRs with respect to the original mesh. For UDF INRs, we measure the Cham- fer Distance (C....

  51. [59]

    These implicit functions are trained using the same source information (meshes)

    Retrieval of INRs trained using Different Source Data In Section 5.4, we demonstrated the retrieval accuracy across different INR implicit functions. These implicit functions are trained using the same source information (meshes). In Tab. 17, we show another case where the UDF...

  52. [60]

    Category-Chamfer Metric 11.1. Retrieval Accuracy by Category and Chamfer Distance Shape retrieval performance is traditionally evaluated based on whether the retrieved shape has the same category as the query shape [37]. While this metric can evaluate the qual- ity of retrieva...

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.