Pith. sign in

REVIEW 3 major objections 5 minor 66 references

Geometric Disentanglement for Generative Latent Shape Models

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read An unsupervised VAE can split the latent space of 3D point clouds into intrinsic shape and articulated pose using only geometry, enabling pose transfer and pose-aware retrieval.

desk verdict Genuinely new Jacobian penalty and honest evaluation, but the intrinsic/extrinsic split is only as good as the near-isometry assumption, which the paper asserts but never validates. read the letter →

arxiv 1908.06386 v1 pith:MQFTAPVO submitted 2019-08-18 cs.CV cs.LGeess.IV

Tristan Aumentado-Armstrong , Stavros Tsogkas , Allan Jepson , Sven Dickinson This is my paper · ORCID
classification cs.CVcs.LGeess.IV
keywords geometricdisentanglementlatentshapemodelsvariationalautoencoderLaplace-Beltramispectrumpointcloudsposetransferpose-awareretrievalhierarchicalpenalties
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

The paper tries to show that a generative latent space for 3D point clouds can be carved into two independent parts with no labels: an 'intrinsic' code that captures body shape or identity, and an 'extrinsic' code that captures articulated pose. The unsupervised target that makes this possible is the Laplace-Beltrami spectrum, a continuous geometric fingerprint that is unchanged by length-preserving deformations. On human and animal datasets, the resulting representation lets a user edit pose while keeping identity fixed, or transfer pose from one subject to another, and supports retrieval by pose or by body shape. If the claim holds, it gives a label-free route to interpretable control of learned 3D shape generators.

What carries the argument

The load-bearing object is the Laplace-Beltrami spectrum $\lambda$, the sorted eigenvalues of the surface Laplacian, which provides a continuous descriptor of intrinsic shape that is invariant to isometric deformations. The GDVAE trains $z_I$ to predict it through a frequency-weighted spectral loss $L_S = \frac{1}{N_\lambda}\sum_{i=1}^{N_\lambda} |\lambda_i - \hat\lambda_i| / i$, where the $1/i$ weighting keeps the low end of the spectrum from being overpowered by high-frequency eigenvalues, as motivated by Weyl's law. The disentanglement penalties are the hierarchical total-correlation decomposition, a hierarchical covariance penalty on inter-group blocks, and the new pairwise Jacobian norm penalty $L_J=\max_{g\neq\tilde g}\|\partial\hat\mu_g/\partial\mu_{\tilde g}\|_F^2$, computed by decoding and re-encoding through the VAE. The Jacobian term directly encodes the geometric requirement that a change in one latent group should not perturb the expected value of another.

What would settle it

Train the same GDVAE on a posed human dataset whose articulations are deliberately coupled to non-isometric surface changes (e.g., BMI varying with joint angle, or a loose cloth simulation), then compare the pose-retrieval error $E_\theta$ obtained with $z_I$ to the entangled-baseline value obtained with $z$. If the near-isometry assumption fails, $E_\theta(z_I)$ will fall toward the baseline while $E_\beta(z_I)$ stays low, showing that pose information has leaked into the intrinsic code.

Watch

Extended reading notes

Core claim

The paper claims that a variational autoencoder for 3D point clouds can learn, without labels, a latent factorization $z=(z_R,z_E,z_I)$ in which $z_R$ controls rigid rotation, $z_E$ controls the extrinsic articulated pose, and $z_I$ controls intrinsic shape identity. In this geometrically disentangled VAE (GDVAE), the intrinsic code is anchored by a spectral loss that forces $z_I$ to predict the Laplace-Beltrami spectrum $\lambda$ of the surface, computed from the training meshes, while the extrinsic and intrinsic codes jointly decode the shape. Three hierarchical penalties enforce the split: the inter-group total-correlation term of a hierarchically factorized VAE, a hierarchical inter-group covariance penalty, and a new pairwise Jacobian norm penalty that measures how much changing one latent group changes the re-encoding of another through the decoder. The paper demonstrates that traversing $z_I$ changes body type or species, traversing $z_E$ changes articulation, swapping $z_E$ transfers pose between subjects, and retrieval using $z_E$ or $z_I$ separately matches the corresponding ground-truth parameters better than an entangled code does.

Load-bearing premise

The method assumes that pose changes are nearly length-preserving (isometric), so the spectral fingerprint it uses as the intrinsic-shape target stays the same across poses of the same subject; if real articulations stretch, squash, or drape the surface, pose information leaks into the intrinsic code.

Editorial extensions

If this is right

  • Pose-aware shape retrieval becomes possible from raw point clouds: querying with $z_E$ matches articulated pose while ignoring identity, and querying with $z_I$ matches identity while ignoring pose.
  • Pose transfer can be done by exchanging $z_E$ between two encoded shapes and decoding, without correspondences, part labels, or mesh connectivity.
  • The latent space supports independent generative control over rotation, pose, and intrinsic shape, so novel samples can be varied in one factor while holding the others fixed.
  • The three penalties are complementary: total correlation reduces all dependence measures, while the direct covariance and Jacobian terms drive their own measures lower, and using all three gives the lowest entanglement values.
  • The paper's retrieval errors provide a quantitative check: using $z_I$ lowers intrinsic-shape error and raises pose error relative to the entangled code, while using $z_E$ does the reverse on the human dataset.

Reading between the lines

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

  • Editorial extension: Because the spectral target is used only during training, a natural stress test is to deploy on pure point clouds with sensor noise and measure whether the retrieval gaps between $z_E$ and $z_I$ persist.
  • Editorial extension: The Jacobian penalty is a general-purpose regularizer for hierarchical VAEs: any pair of latent blocks that should be causally independent could be penalized the same way, with no geometric interpretation required.
  • Editorial extension: If real pose variation is non-isometric, the factorization could be enriched by adding a pose-conditioned correction to the spectral predictor, forcing $z_I$ to drop pose information even when the geometry target leaks it.
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

3 major / 5 minor

Summary. The paper proposes GDVAE, a two-level generative model for 3D point clouds that factorizes the latent space of a VAE into three groups: rotation zR, extrinsic pose zE, and intrinsic shape zI. The factorization is driven by a spectral loss that trains zI to predict the Laplace-Beltrami (LBO) spectrum of the input mesh, together with hierarchical disentanglement penalties (total correlation, inter-group covariance, and a novel Jacobian-based penalty). The model is evaluated on MNIST height fields, Dyna, SMAL, and SMPL datasets through reconstruction, generative sampling, latent interpolation, pose transfer, and pose-aware retrieval. The central claim is that an unsupervised, geometry-only objective yields an interpretable intrinsic/extrinsic split that enables tasks such as pose transfer and pose-aware shape retrieval.

Significance. If fully validated, the paper would make a useful contribution: using the LBO spectrum as an unsupervised training signal for latent shape factorization is an appealing idea, and the proposed Jacobian penalty is a reasonable addition to the hierarchical disentanglement toolbox. The paper also provides extensive ablations and a retrieval protocol with ground-truth shape/pose parameters on synthetic data, which is valuable for future comparisons. However, the quantitative support for the central claim is mixed: the near-isometry assumption underlying the spectral objective is asserted rather than measured, and the retrieval results do not consistently show that the extrinsic subgroup outperforms the entangled baseline. The paper's own pose-transfer evaluation acknowledges visible failures, which tempers the claim of a clean factorization.

major comments (3)
  1. [Appendix B / §4.2] The entire intrinsic/extrinsic split rests on the claim that articulations are nearly isometric, so that the LBO spectrum (the target of Eq. (7)) is invariant to pose. This is asserted in Appendix B but never quantified. SMPL and SMAL are generated by linear blend skinning with sampled joint angles; skinning produces local stretching and compression around joints, which changes the surface metric and hence the spectrum. If pose changes the spectrum, then the spectral loss forces zI to encode pose information, and the covariance/Jacobian penalties cannot remove that information without sacrificing spectral prediction. The retrieval results are consistent with this concern: on SMAL, Eθ using zE equals the entangled baseline (0.983 vs 0.983, Table 2), and on SMPL the improvement is small (0.709 vs 0.726). Please provide a quantitative validation of the isometry assumption, for example the distribution of spectrum distances across poses of the same subject versus across subjects, or demonstrate that the predicted spectrum from zI is insensitive to pose. This is a load-bearing point for the paper's central claim.
  2. [§5.3, Table 2] The pose-aware retrieval results do not demonstrate that zE is better than the entangled baseline. On SMAL, Eθ(zE)=0.983 is identical to Eθ(z)=0.983; on SMPL, Eθ(zE)=0.709 versus Eθ(z)=0.726 is a small difference that is comparable to the reported SEM values in Table 6 (up to 0.0058 across model runs and 0.0070 across shape samplings). Moreover, the paper's description of 'much lower' errors is not borne out by the magnitudes: on SMAL, Eθ(zE)=0.983 versus Eθ(zI)=0.993 is a difference of only 0.01. The authors should report confidence intervals or a paired significance test, and should explicitly discuss the SMAL null result. Without a decisive margin over the entangled representation, the claim that the model enables pose-aware retrieval is not established.
  3. [§5.2] The paper's own qualitative evaluation reports that pose transfer on SMPL and Dyna fails in the sense that 'the transferred arm positions tend to be similar, but not exactly the same. This suggests a failure in the disentanglement, since the articulations are tied to the latent intrinsics zI.' This statement directly contradicts the abstract's claim that the representation 'exhibits intuitive and interpretable behavior, enabling tasks such as pose transfer.' Either the central claim needs to be tempered to an approximate factorization, or the authors should provide a quantitative pose-transfer metric (e.g., joint-angle error between the transferred shape and the target pose) to characterize the degree of failure. As written, the paper's own evidence indicates that the split is not clean on the datasets used for the main demonstration.
minor comments (5)
  1. [Table 1] Please ensure that the 'z S' header in Table 1 is clearly separated into 'z' and 'S' columns, since the current formatting is ambiguous and the reader may miscount the columns.
  2. [Eq. (4)] The KL divergence DKL is used without a prior definition; define qφ and p explicitly or cite the standard VAE formulation so that the hierarchical decomposition is self-contained.
  3. [§5.3] The normalization of the retrieval errors by 'the average error between all shape pairs' is described only in a sentence; clarify the exact normalization factor and report unnormalized values or a random baseline for interpretability.
  4. [Appendix C.2 / Table 3] The choices of β4, γI, and wJ across datasets are presented without justification; a brief sensitivity discussion in the main text would help, given that these weights are central to the method.
  5. [Figure 3] The caption refers to 'red and blue dashed paths', but the figure appears to use grayscale rendering; update the caption or the figure colors to match.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the spectral loss is a self-supervised training signal, and the disentanglement claims are validated against external SMPL/SMAL parameters.

full rationale

The paper's derivation chain is self-contained and non-circular. The intrinsic/extrinsic split is operationalized by training the latent subgroup zI to predict the Laplace–Beltrami spectrum (Eq. 7, with λ̂ = fS(zI)), while the hierarchical penalties (LCOV in Eq. 5, LJ in Eq. 6, and the HFVAE total-correlation term) penalize statistical dependence between zI and zE. This is a self-supervised learning objective, not a tautology: zI is not defined as the spectrum; it is a latent variable optimized to encode it, and the disentanglement claims are evaluated against external SMPL/SMAL body-shape and pose parameters β and θ that were never used in training. The paper explicitly reports partial failures, e.g., Section 5.2: 'the transferred arm positions tend to be similar, but not exactly the same. This suggests a failure in the disentanglement, since the articulations are tied to the latent intrinsics zI.' Such an admission is inconsistent with a result forced by construction. No load-bearing step reduces to its own inputs; the near-isometry premise in Appendix B is an empirical assumption about the datasets, not a circular redefinition of the target.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claim rests mainly on domain assumptions about isometry and spectrum completeness, plus hand-chosen loss weights. No fitted constants are used to produce the retrieval metrics; those use hold-out ground truth. The balance of assumptions is standard for a deep learning method paper.

free parameters (4)
  • Inter-group TC weight beta4 = 50 (MNIST, SMAL, SMPL); 25 (Dyna)
    Tuned per dataset; controls strength of information-theoretic disentanglement; high values hurt reconstruction (Sec. F.1).
  • Covariance penalty weight gamma_I = 1 (MNIST); 5 (Dyna); 100 (SMAL); 10 (SMPL)
    Set by hand per dataset; ablations show it is needed to keep inter-group covariance small.
  • Jacobian penalty weight w_J = 1 (MNIST); 5 (Dyna); 10 (SMAL, SMPL)
    Chosen by hand; weights the novel Jacobian penalty term in Eq. 6.
  • Spectral loss weight zeta = 1000 (MNIST, Dyna, SMAL); 500 (SMPL)
    Weights the LBO spectrum prediction loss in Eq. 9; large values force zI to encode the spectrum.
assumptions (5)
  • domain assumption Articulated pose changes in the data are near-isometric, making the LBO spectrum pose-invariant
    Invoked in Appendix B ('articulations... are nearly isometric transformations'); if false, the spectral target cannot separate pose from identity.
  • domain assumption A 3D shape can be modeled as a 2D Riemannian manifold embedded in 3D, with point clouds as random surface samplings
    Sec. 4.2 uses this to define the LBO spectrum from meshes; real scanned point clouds with noise and outliers deviate.
  • ad hoc to paper The first Nlambda eigenvalues of the LBO spectrum sufficiently characterize intrinsic shape
    The spectral loss compares truncated spectra with Nlambda = 40 to 100 (Sec. 5, Appendix A); truncation discards high-frequency detail.
  • standard math Weyl's law justifies the linear weighting 1/i in the spectral loss
    Sec. 4.2 cites Weyl's law for approximately linear growth of eigenvalues; used only to design the loss, not the core claim.
  • ad hoc to paper The AE's factorization into a rotation quaternion R and shape vector X is learnable and separates rigid rotation from the rest
    The two-level architecture (Sec. 3.3, Fig. 2) assumes the decoder can apply rotation as a matrix multiplication on a canonical shape representation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Geometric Disentanglement for Generative Latent Shape Models." pith.science (2026). https://pith.science/paper/MQFTAPVO

@misc{pith2026190806386,
  author       = {Pith},
  title        = {Pith review of: Geometric Disentanglement for Generative Latent Shape Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MQFTAPVO}},
  note         = {Machine review of arXiv:1908.06386}
}
read the original abstract

Representing 3D shape is a fundamental problem in artificial intelligence, which has numerous applications within computer vision and graphics. One avenue that has recently begun to be explored is the use of latent representations of generative models. However, it remains an open problem to learn a generative model of shape that is interpretable and easily manipulated, particularly in the absence of supervised labels. In this paper, we propose an unsupervised approach to partitioning the latent space of a variational autoencoder for 3D point clouds in a natural way, using only geometric information. Our method makes use of tools from spectral differential geometry to separate intrinsic and extrinsic shape information, and then considers several hierarchical disentanglement penalties for dividing the latent space in this manner, including a novel one that penalizes the Jacobian of the latent representation of the decoded output with respect to the latent encoding. We show that the resulting representation exhibits intuitive and interpretable behavior, enabling tasks such as pose transfer and pose-aware shape retrieval that cannot easily be performed by models with an entangled representation.

Figures

Figures reproduced from arXiv: 1908.06386 by the authors.

Figure 1
Figure 1. Factoring pose and intrinsic shape within a disentangled [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. A schematic overview of the combined two-level ar [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Diagram of the pairwise Jacobian norm penalty com [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (14 more)
Figure 4
Figure 4. Figure 4: Reconstructions of random samples, passed through both the AE and VAE. For each pair, the left shape is the input and the right shape is the reconstruction. Colors denote depth (i.e., distance from the camera). Rows: MNIST, Dyna, SMAL, SMPL. where LQ is the metric over…
Figure 5
Figure 5. Figure 5: Samples drawn from the latent space of the VAE by decoding z ∼ N (0, I) with zR = 0. Colors denote depth (i.e., distance from the camera). Rows: MNIST, Dyna, SMAL, SMPL. zR zE zI zRE zRI zEI z S 0.32 0.47 0.60 0.64 0.68 0.88 0.88 0.98 [PITH_FULL_IMAGE:figures/full_fig…
Figure 6
Figure 6. Figure 6: Latent space interpolations between SMPL (row 1) and SMAL (row 2) shapes. Each inset interpolates z between the upper-left and lower right shapes, with zE changing along the vertical axis and zI changing along the horizontal one. Per-shape colours denote depth [PITH_F…
Figure 7
Figure 7. Figure 7: Pose transfer via exchanging latent extrinsics. Per inset of four shapes, the bottom shapes have the zR and zI of the shape directly above, but the zE of their diagonally opposite shape in the top row. Per-shape colors denote depth. Upper shapes are real point clouds; …
Figure 8
Figure 8. Figure 8: Effect of randomly sampling either the intrinsic or extrinsic components of four Dyna shapes. Leftmost shape: original input; upper row: zI ∼ N (0, I), fixed zE ; lower row: zE ∼ N (0, I), fixed zI . Colors denote depth (distance from the camera). X z zE zI SMAL Eβ 0.6…
Figure 9
Figure 9. Figure 9: Shape retrieval. Per inset: leftmost shape is query, mid￾dle two shapes are retrieved via zE, and rightmost two shapes are retrieved via zI . Color gradients per shape denote depth. retrieval [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 10
Figure 10. Figure 10: Confusion matrix of a shallow classifier, mapping mesh [PITH_FULL_IMAGE:figures/full_fig_p012_10.png]
Figure 11
Figure 11. Figure 11: A t-SNE plot of the LBO spectra of the human shapes in [PITH_FULL_IMAGE:figures/full_fig_p013_11.png]
Figure 12
Figure 12. Figure 12: Disentangled latent interpolations for the MNIST dataset. Colours per digit denote depth. For each inset, we are interpolating [PITH_FULL_IMAGE:figures/full_fig_p015_12.png]
Figure 13
Figure 13. Figure 13: Empirical curves of loss terms during training across weight hyper-parameters on model learning. See Table [PITH_FULL_IMAGE:figures/full_fig_p015_13.png]
Figure 14
Figure 14. Figure 14: Confusion matrices for a linear SVM classifier accuracy on the MNIST test set. Top row: testing with [PITH_FULL_IMAGE:figures/full_fig_p016_14.png]
Figure 15
Figure 15. Figure 15: Examples of autoencoder reconstructions across different rotations of the same object. Note that the encoding of the datum from [PITH_FULL_IMAGE:figures/full_fig_p017_15.png]
Figure 17
Figure 17. Figure 17: Empirical curves of loss terms for disentanglement ablations on SMAL. NC, NJ, and NJC mean no covariance, no Jacobian, [PITH_FULL_IMAGE:figures/full_fig_p018_17.png]
Figure 18
Figure 18. Figure 18: Empirical curves of loss terms for disentanglement ablations on SMPL. NC, NJ, and NJC mean no covariance, no Jacobian, or [PITH_FULL_IMAGE:figures/full_fig_p019_18.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

66 extracted references · 37 canonical work pages

  1. [1]

    Learning representations and gen- erative models for 3d point clouds

    Panos Achlioptas, Olga Diamanti, Ioannis Mitliagkas, and Leonidas Guibas. Learning representations and gen- erative models for 3d point clouds. arXiv preprint arXiv:1707.02392, 2017. 2, 3, 6, 13, 14

    arXiv 2017

  2. [2]

    Hyperprior induced un- supervised disentanglement of latent representations

    Abdul Fatir Ansari and Harold Soh. Hyperprior induced un- supervised disentanglement of latent representations. arXiv preprint arXiv:1809.04497, 2018. 2

    arXiv 2018

  3. [3]

    Point convolutional neural networks by extension operators

    Matan Atzmon, Haggai Maron, and Yaron Lipman. Point convolutional neural networks by extension operators. arXiv preprint arXiv:1803.10091, 2018. 2

    arXiv 2018

  4. [4]

    Layer normalization

    Jimmy Lei Ba, Jamie Ryan Kiros, and Geoffrey E Hin- ton. Layer normalization. arXiv preprint arXiv:1607.06450,

    arXiv

  5. [5]

    Possible principles underlying the transformation of sensory messages

    Horace B Barlow et al. Possible principles underlying the transformation of sensory messages. Sensory communica- tion, 1:217–234, 1961. 1

    work page 1961

  6. [6]

    Constructing laplace operator from point clouds in Rd

    Mikhail Belkin, Jian Sun, and Yusu Wang. Constructing laplace operator from point clouds in Rd. In Proceedings of the twentieth annual ACM-SIAM symposium on Discrete algorithms, pages 1031–1040. Society for Industrial and Ap- plied Mathematics, 2009. 5, 20

    work page 2009

  7. [7]

    Rep- resentation learning: A review and new perspectives

    Yoshua Bengio, Aaron Courville, and Pascal Vincent. Rep- resentation learning: A review and new perspectives. IEEE transactions on pattern analysis and machine intelligence , 35(8):1798–1828, 2013. 1

    2013

  8. [8]

    Multi-level variational autoencoder: Learning disentangled representations from grouped observations

    Diane Bouchacourt, Ryota Tomioka, and Sebastian Nowozin. Multi-level variational autoencoder: Learning disentangled representations from grouped observations. arXiv preprint arXiv:1705.08841, 2017. 2

    arXiv 2017

Show all 66 references
  1. [9]

    Shape google: Geometric words and expressions for invariant shape retrieval

    Alexander M Bronstein, Michael M Bronstein, Leonidas J Guibas, and Maks Ovsjanikov. Shape google: Geometric words and expressions for invariant shape retrieval. ACM Transactions on Graphics (TOG), 30(1):1, 2011. 6

    2011

  2. [10]

    Understanding disentangling in beta-vae

    Christopher P Burgess, Irina Higgins, Arka Pal, Loic Matthey, Nick Watters, Guillaume Desjardins, and Alexan- der Lerchner. Understanding disentangling in beta-vae. arXiv preprint arXiv:1804.03599, 2018. 2

    2018 arXiv

  3. [11]

    Isolating sources of disentanglement in variational autoencoders

    Tian Qi Chen, Xuechen Li, Roger Grosse, and David Du- venaud. Isolating sources of disentanglement in variational autoencoders. arXiv preprint arXiv:1802.04942, 2018. 2

    2018 arXiv

  4. [12]

    Infogan: Interpretable repre- sentation learning by information maximizing generative ad- versarial nets

    Xi Chen, Yan Duan, Rein Houthooft, John Schulman, Ilya Sutskever, and Pieter Abbeel. Infogan: Interpretable repre- sentation learning by information maximizing generative ad- versarial nets. In Advances in neural information processing systems, pages 2172–2180, 2016. 2

    2016

  5. [13]

    Functional char- acterization of intrinsic and extrinsic geometry

    Etienne Corman, Justin Solomon, Mirela Ben-Chen, Leonidas Guibas, and Maks Ovsjanikov. Functional char- acterization of intrinsic and extrinsic geometry. ACM Trans- actions on Graphics (TOG), 36(2):14, 2017. 4

    2017

  6. [14]

    Learning to sam- ple

    Oren Dovrat, Itai Lang, and Shai Avidan. Learning to sam- ple. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 2760–2769, 2019. 3

    2019

  7. [15]

    Structured dis- entangled representations

    Babak Esmaeili, Hao Wu, Sarthak Jain, Alican Bozkurt, Narayanaswamy Siddharth, Brooks Paige, Dana H Brooks, Jennifer Dy, and Jan-Willem van de Meent. Structured dis- entangled representations. arXiv preprint arXiv:1804.02086,

    arXiv

  8. [16]

    Multi-view face detection using deep convolutional neural networks

    Sachin Sudhakar Farfade, Mohammad J Saberian, and Li- Jia Li. Multi-view face detection using deep convolutional neural networks. In Proceedings of the 5th ACM on Interna- tional Conference on Multimedia Retrieval, pages 643–650. ACM, 2015. 3

    2015

  9. [17]

    Automatic unpaired shape deformation transfer

    Lin Gao, Jie Yang, Yi-Ling Qiao, Yu-Kun Lai, Paul L Rosin, Weiwei Xu, and Shihong Xia. Automatic unpaired shape deformation transfer. In SIGGRAPH Asia 2018 Technical Papers, page 237. ACM, 2018. 2

    2018

  10. [18]

    Auto-encoding total correlation explanation.arXiv preprint arXiv:1802.05822, 2018

    Shuyang Gao, Rob Brekelmans, Greg Ver Steeg, and Aram Galstyan. Auto-encoding total correlation explanation.arXiv preprint arXiv:1802.05822, 2018. 2

    2018 arXiv

  11. [19]

    Generative adversarial nets

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial nets. In Advances in neural information processing systems, pages 2672–2680,

  12. [20]

    Atlasnet: A papier-mache ap- proach to learning 3d surface generation

    Thibault Groueix, Matthew Fisher, Vladimir Kim, Bryan Russell, and Mathieu Aubry. Atlasnet: A papier-mache ap- proach to learning 3d surface generation. In CVPR 2018 ,

    2018

  13. [21]

    3d-coded: 3d cor- respondences by deep deformation

    Thibault Groueix, Matthew Fisher, Vladimir G Kim, Bryan C Russell, and Mathieu Aubry. 3d-coded: 3d cor- respondences by deep deformation. In Proceedings of the European Conference on Computer Vision (ECCV) , pages 230–246, 2018. 12

    2018

  14. [22]

    A two-step disentanglement method

    Naama Hadad, Lior Wolf, and Moni Shahar. A two-step disentanglement method. In Proceedings of the IEEE Con- ference on Computer Vision and Pattern Recognition, pages 772–780, 2018. 2

    2018

  15. [23]

    Disentangling latent fac- tors with whitening

    Sangchul Hahn and Heeyoul Choi. Disentangling latent fac- tors with whitening. arXiv preprint arXiv:1811.03444, 2018. 2

    2018 arXiv

  16. [24]

    Monte carlo convolution for learning on non-uniformly sampled point clouds

    Pedro Hermosilla, Tobias Ritschel, Pere-Pau V ´azquez, `Alvar Vinacua, and Timo Ropinski. Monte carlo convolution for learning on non-uniformly sampled point clouds. arXiv preprint arXiv:1806.01759, 2018. 2

    2018 arXiv

  17. [25]

    Early visual concept learning with unsuper- vised deep learning

    Irina Higgins, Loic Matthey, Xavier Glorot, Arka Pal, Be- nigno Uria, Charles Blundell, Shakir Mohamed, and Alexan- der Lerchner. Early visual concept learning with unsuper- vised deep learning. arXiv preprint arXiv:1606.05579, 2016. 1

    2016 arXiv

  18. [26]

    β-vae: Learning basic visual concepts with a constrained variational framework

    Irina Higgins, Loic Matthey, Arka Pal, Christopher Burgess, Xavier Glorot, Matthew Botvinick, Shakir Mohamed, and Alexander Lerchner. β-vae: Learning basic visual concepts with a constrained variational framework. In International Conference on Learning Representations, 2017. 2, 17

    2017

  19. [27]

    The role of independent motion in object segmentation in the ventral visual stream: Learning to recognise the separate parts of the body

    IV Higgins and SM Stringer. The role of independent motion in object segmentation in the ventral visual stream: Learning to recognise the separate parts of the body. Vision research, 51(6):553–562, 2011. 1

    2011

  20. [28]

    A simple probabilistic deep generative model for learning generalizable disentangled representations from grouped data

    Haruo Hosoya. A simple probabilistic deep generative model for learning generalizable disentangled representations from grouped data. arXiv preprint arXiv:1809.02383, 2018. 2

    2018 arXiv

  21. [29]

    Metrics for 3d rotations: Comparison and analysis

    Du Q Huynh. Metrics for 3d rotations: Comparison and analysis. Journal of Mathematical Imaging and Vision , 35(2):155–164, 2009. 3

    2009

  22. [30]

    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. arXiv preprint arXiv:1502.03167, 2015. 13

    2015 arXiv

  23. [31]

    Rotation invariant spherical harmonic repre- sentation of 3d shape descriptors

    Michael Kazhdan, Thomas Funkhouser, and Szymon Rusinkiewicz. Rotation invariant spherical harmonic repre- sentation of 3d shape descriptors. InSymposium on geometry processing, volume 6, pages 156–164, 2003. 3

    2003

  24. [32]

    Disentangling by factoris- ing

    Hyunjik Kim and Andriy Mnih. Disentangling by factoris- ing. arXiv preprint arXiv:1802.05983, 2018. 2

    2018 arXiv

  25. [33]

    Adam: A method for stochastic optimization

    Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980 ,

    arXiv

  26. [34]

    Auto-encoding varia- tional bayes

    Diederik P Kingma and Max Welling. Auto-encoding varia- tional bayes. arXiv preprint arXiv:1312.6114, 2013. 1

    2013 arXiv

  27. [35]

    Variational inference of disentangled latent concepts from unlabeled observations

    Abhishek Kumar, Prasanna Sattigeri, and Avinash Bal- akrishnan. Variational inference of disentangled latent concepts from unlabeled observations. arXiv preprint arXiv:1711.00848, 2017. 2, 4

    2017 arXiv

  28. [36]

    Autoencoding beyond pix- els using a learned similarity metric

    Anders Boesen Lindbo Larsen, Søren Kaae Sønderby, Hugo Larochelle, and Ole Winther. Autoencoding beyond pix- els using a learned similarity metric. arXiv preprint arXiv:1512.09300, 2015. 6

    2015 arXiv

  29. [37]

    Gradient-based learning applied to document recog- nition

    Yann LeCun, L ´eon Bottou, Yoshua Bengio, and Patrick Haffner. Gradient-based learning applied to document recog- nition. Proceedings of the IEEE , 86(11):2278–2324, 1998. 5

    1998

  30. [38]

    Point cloud gan

    Chun-Liang Li, Manzil Zaheer, Yang Zhang, Barnabas Poc- zos, and Ruslan Salakhutdinov. Point cloud gan. arXiv preprint arXiv:1810.05795, 2018. 2, 3

    2018 arXiv

  31. [39]

    So-net: Self- organizing network for point cloud analysis

    Jiaxin Li, Ben M Chen, and Gim Hee Lee. So-net: Self- organizing network for point cloud analysis. In Proceed- ings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 9397–9406, 2018. 3

    2018

  32. [40]

    Matthew Loper, Naureen Mahmood, Javier Romero, Ger- ard Pons-Moll, and Michael J. Black. SMPL: A skinned multi-person linear model. ACM Trans. Graphics (Proc. SIGGRAPH Asia), 34(6):248:1–248:16, Oct. 2015. 5, 12

    2015

  33. [41]

    Visualiz- ing data using t-sne

    Laurens van der Maaten and Geoffrey Hinton. Visualiz- ing data using t-sne. Journal of machine learning research, 9(Nov):2579–2605, 2008. 13

    2008

  34. [42]

    Adversarial variational bayes: Unifying variational autoen- coders and generative adversarial networks

    Lars Mescheder, Sebastian Nowozin, and Andreas Geiger. Adversarial variational bayes: Unifying variational autoen- coders and generative adversarial networks. In Proceedings of the 34th International Conference on Machine Learning- Volume 70, pages 2391–2400. JMLR. org, 2017. 2

    2017

  35. [43]

    Discrete differential-geometry operators for triangu- lated 2-manifolds

    Mark Meyer, Mathieu Desbrun, Peter Schr ¨oder, and Alan H Barr. Discrete differential-geometry operators for triangu- lated 2-manifolds. In Visualization and mathematics III , pages 35–57. Springer, 2003. 5

    2003

  36. [44]

    Learning disentangled representations with semi-supervised deep generative mod- els

    Siddharth Narayanaswamy, Brooks Paige, Jan-Willem Van de Meent, Alban Desmaison, Noah Goodman, Pushmeet Kohli, Frank Wood, and Philip Torr. Learning disentangled representations with semi-supervised deep generative mod- els. In Advances in Neural Information Processing Systems...

    2017

  37. [45]

    The shape variational autoencoder: A deep generative model of part-segmented 3d objects

    Charlie Nash and Chris KI Williams. The shape variational autoencoder: A deep generative model of part-segmented 3d objects. In Computer Graphics Forum , volume 36, pages 1–12. Wiley Online Library, 2017. 2

    2017

  38. [46]

    Automatic differentiation in pytorch

    Adam Paszke, Sam Gross, Soumith Chintala, Gregory Chanan, Edward Yang, Zachary DeVito, Zeming Lin, Al- ban Desmaison, Luca Antiga, and Adam Lerer. Automatic differentiation in pytorch. 2017. 13

    2017

  39. [47]

    Scikit-learn: Machine learning in python

    Fabian Pedregosa, Gael Varoquaux, Alexandre Gramfort, Vincent Michel, Bertrand Thirion, Olivier Grisel, Mathieu Blondel, Peter Prettenhofer, Ron Weiss, Vincent Dubourg, Jake VanderPlas, Alexandre Passos, David Cournapeau, Matthieu Brucher, Matthieu Perrot, and Edouard Duches- ...

    2012 arXiv

  40. [48]

    Shape retrieval of non-rigid 3d human models.International Journal of Computer Vision, 120(2):169–193, 2016

    David Pickup, Xianfang Sun, Paul L Rosin, Ralph R Martin, Z Cheng, Zhouhui Lian, Masaki Aono, A Ben Hamza, A Bronstein, M Bronstein, et al. Shape retrieval of non-rigid 3d human models.International Journal of Computer Vision, 120(2):169–193, 2016. 6

    2016

  41. [49]

    Gerard Pons-Moll, Javier Romero, Naureen Mahmood, and Michael J. Black. Dyna: A model of dynamic human shape in motion. ACM Transactions on Graphics, (Proc. SIG- GRAPH), 34(4):120:1–120:14, Aug. 2015. 5, 12

    2015

  42. [50]

    Pointnet: Deep learning on point sets for 3d classifica- tion and segmentation

    Charles R Qi, Hao Su, Kaichun Mo, and Leonidas J Guibas. Pointnet: Deep learning on point sets for 3d classifica- tion and segmentation. Proc. Computer Vision and Pattern Recognition (CVPR), IEEE, 1(2):4, 2017. 2, 3, 13

    2017

  43. [51]

    Pointnet++: Deep hierarchical feature learning on point sets in a metric space

    Charles Ruizhongtai Qi, Li Yi, Hao Su, and Leonidas J Guibas. Pointnet++: Deep hierarchical feature learning on point sets in a metric space. In Advances in Neural Informa- tion Processing Systems, pages 5099–5108, 2017. 2

    2017

  44. [52]

    Deep learning with sets and point clouds

    Siamak Ravanbakhsh, Jeff Schneider, and Barnabas Poczos. Deep learning with sets and point clouds. arXiv preprint arXiv:1611.04500, 2016. 2

    2016 arXiv

  45. [53]

    Laplace–beltrami spectra as shape-dna of surfaces and solids

    Martin Reuter, Franz-Erich Wolter, and Niklas Peinecke. Laplace–beltrami spectra as shape-dna of surfaces and solids. Computer-Aided Design, 38(4):342–366, 2006. 5

    2006

  46. [54]

    Stochastic backpropagation and approximate inference in deep generative models

    Danilo Jimenez Rezende, Shakir Mohamed, and Daan Wier- stra. Stochastic backpropagation and approximate inference in deep generative models. arXiv preprint arXiv:1401.4082,

    arXiv

  47. [55]

    Learning Disentangled Representations with Reference-Based Variational Autoencoders

    Adri `a Ruiz, Oriol Martinez, Xavier Binefa, and Jakob Verbeek. Learning Disentangled Representations with Reference-Based Variational Autoencoders. working paper or preprint, Oct. 2018. 2

    2018

  48. [56]

    Vari- ational autoencoders for deforming 3d mesh models

    Qingyang Tan, Lin Gao, Yu-Kun Lai, and Shihong Xia. Vari- ational autoencoders for deforming 3d mesh models. In Pro- ceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 5841–5850, 2018. 2

    2018

  49. [57]

    Vae with a vampprior

    Jakub M Tomczak and Max Welling. Vae with a vampprior. arXiv preprint arXiv:1705.07120, 2017. 6

    2017 arXiv

  50. [58]

    Spectral geometry process- ing with manifold harmonics

    Bruno Vallet and Bruno L ´evy. Spectral geometry process- ing with manifold harmonics. In Computer Graphics Forum, volume 27, pages 251–260. Wiley Online Library, 2008. 5

    2008

  51. [59]

    Learn- ing localized generative models for 3d point clouds via graph convolution

    Diego Valsesia, Giulia Fracastoro, and Enrico Magli. Learn- ing localized generative models for 3d point clouds via graph convolution. In International Conference on Learning Rep- resentations, 2019. 2

    2019

  52. [60]

    Black, Ivan Laptev, and Cordelia Schmid

    G ¨ul Varol, Javier Romero, Xavier Martin, Naureen Mah- mood, Michael J. Black, Ivan Laptev, and Cordelia Schmid. Learning from synthetic humans. In CVPR, 2017. 5, 12

    2017

  53. [61]

    Information theoretical analysis of multi- variate correlation

    Satosi Watanabe. Information theoretical analysis of multi- variate correlation. IBM Journal of research and develop- ment, 4(1):66–82, 1960. 2

    1960

  54. [62]

    Spidercnn: Deep learning on point sets with parameter- ized convolutional filters

    Yifan Xu, Tianqi Fan, Mingye Xu, Long Zeng, and Yu Qiao. Spidercnn: Deep learning on point sets with parameter- ized convolutional filters. arXiv preprint arXiv:1803.11527,

    arXiv

  55. [63]

    Fold- ingnet: Point cloud auto-encoder via deep grid deformation

    Yaoqing Yang, Chen Feng, Yiru Shen, and Dong Tian. Fold- ingnet: Point cloud auto-encoder via deep grid deformation. In Proc. IEEE Conf. on Computer Vision and Pattern Recog- nition (CVPR), volume 3, 2018. 3

    2018

  56. [64]

    The information autoencoding family: A lagrangian perspec- tive on latent variable generative models

    Shengjia Zhao, Jiaming Song, and Stefano Ermon. The information autoencoding family: A lagrangian perspec- tive on latent variable generative models. arXiv preprint arXiv:1806.06514, 2018. 2

    2018 arXiv

  57. [65]

    T-networks

    Silvia Zuffi, Angjoo Kanazawa, David Jacobs, and Michael J. Black. 3D menagerie: Modeling the 3D shape and pose of animals. In IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), July 2017. 5, 12 Appendix A. Dataset Details A.1. MNIST Dataset As a simple dataset on wh...

    2017

  58. [66]

    -R”), or using an LBO estimated from a point cloud (denoted “-P

    Recall that low Eθ using zE and low Eβ using zI are good (indicating zI and zE hold intrinsic shape and pose respectively), while lowEθ usingzI and lowEβ usingzE are not (as it means entanglement is present). For SMAL, we see that REG has the best Eβ using zI, while NJC is wor...

    2000

Pith tools

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