mlsanityData to 7 Oct 2026

3D Shape Variational Autoencoder Latent Disentanglement via Mini-Batch Feature Swapping for Bodies and Faces

arXiv 2111.12448 ↗First posted 24 Nov 2021cs.CV cs.GR cs.LGcited by 15

Not on the map. Why?

Abstract (arXiv)

Learning a disentangled, interpretable, and structured latent representation in 3D generative models of faces and bodies is still an open problem. The problem is particularly acute when control over identity features is required. In this paper, we propose an intuitive yet effective self-supervised approach to train a 3D shape variational autoencoder (VAE) which encourages a disentangled latent representation of identity features. Curating the mini-batch generation by swapping arbitrary features across different shapes allows to define a loss function leveraging known differences and similarities in the latent representations. Experimental results conducted on 3D meshes show that state-of-the-art methods for latent disentanglement are not able to disentangle identity features of faces and bodies. Our proposed method properly decouples the generation of such features while maintaining good representation and reconstruction capabilities.

Key facts. Cited by 15 papers in the map (data through 7 Oct 2026). Builds on 30 of its 47 references that are in the map. First posted on arXiv 24 Nov 2021.

Builds on

30 of 47 references.

References · 30

  1. 2014Adam: A Method for Stochastic Optimizationcited by 36,960
  2. 2013Auto-Encoding Variational Bayescited by 3,980
  3. 2018Dynamic Graph CNN for Learning on Point Cloudscited by 2,220
  4. 2018Disentangling by Factorisingcited by 614
  5. 2019PointFlow: 3D Point Cloud Generation with Continuous Normalizing Flowscited by 487
  6. 2017Learning from Synthetic Humanscited by 358
  7. 20163D Menagerie: Modeling the 3D shape and pose of animalscited by 310
  8. 20193D Morphable Face Models -- Past, Present and Futurecited by 269
  9. 2018Generating 3D faces using Convolutional Mesh Autoencoderscited by 267
  10. 2017Variational Inference of Disentangled Latent Concepts from Unlabeled Observationscited by 240
  11. 2020STAR: Sparse Trained Articulated Human Body Regressorcited by 194
  12. 2019Learning to Regress 3D Face Shape and Expression from an Image without 3D Supervisioncited by 176
18 more
  1. 2015Deep Convolutional Inverse Graphics Networkcited by 121
  2. 2020Learning Formation of Physically-Based Face Attributescited by 94
  3. 2019SpiralNet++: A Fast and Highly Efficient Mesh Convolution Operatorcited by 89
  4. 2018Variational Autoencoders Pursue PCA Directions (by Accident)cited by 84
  5. 2020Measuring Disentanglement: A Review of Metricscited by 69
  6. 2019Towards a complete 3D morphable model of the human headcited by 63
  7. 2018Structured Disentangled Representationscited by 62
  8. 2020Guided Variational Autoencoder for Disentanglement Learningcited by 57
  9. 2020LIMP: Learning Latent Shape Representations with Metric Preservation Priorscited by 54
  10. 2020Fully Convolutional Mesh Autoencoder using Efficient Spatially Varying Kernelscited by 53
  11. 2017Deformable Shape Completion with Graph Convolutional Autoencoderscited by 50
  12. 2019Combining 3D Morphable Models: A Large scale Face-and-Head Modelcited by 49
  13. 2019Unsupervised Model Selection for Variational Disentangled Representation Learningcited by 46
  14. 2019MeshGAN: Non-linear 3D Morphable Models of Facescited by 39
  15. 2019Synthesizing Coupled 3D Face Modalities by Trunk-Branch Generative Adversarial Networkscited by 34
  16. 2019Geometric Disentanglement for Generative Latent Shape Modelscited by 30
  17. 2019Mesh Variational Autoencoders with Edge Contraction Poolingcited by 14
  18. 2020Intraoperative Liver Surface Completion with Graph Convolutional VAEcited by 9

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 6; 2024: 3; 2025: 4; 2026: 2.

15 papers.

Sort:

By area:

  1. 2025Taming Video Models for 3D and 4D Generation via Zero-Shot Camera Controlcited by 14
  2. 2023TUVF: Learning Generalizable Texture UV Radiance Fieldscited by 9
  3. 20233D Generative Model Latent Disentanglement via Local Eigenprojectioncited by 6
  4. 2024ShapeFusion: A 3D diffusion model for localized shape editingcited by 6
  5. 2024Locally Adaptive Neural 3D Morphable Modelscited by 5
  6. 2025SMPL-GPTexture: Dual-View 3D Human Texture Estimation using Text-to-Image Generation Modelscited by 2
  7. 2025BFSM: 3D Bidirectional Face-Skull Morphable Modelcited by 2
  8. 2026Parallelised Differentiable Straightest Geodesics for 3D Meshescited by 2
  9. 2023Unsupervised Representation Learning for Diverse Deformable Shape Collectionscited by 1
  10. 2024Disentangling Hippocampal Shape Variations: A Study of Neurological Disorders Using Mesh Variational Autoencoder with Contrastive Learningcited by 1
  11. 2023Latent Disentanglement in Mesh Variational Autoencoders Improves the Diagnosis of Craniofacial Syndromes and Aids Surgical Planningcited by 0
  12. 2023MUNCH: Modelling Unique 'N Controllable Headscited by 0
3 more
  1. 2023Plug-and-Play Feature Generation for Few-Shot Medical Image Classificationcited by 0
  2. 2025Conditional Diffusion Model with Anatomical-Dose Dual Constraints for End-to-End Multi-Tumor Dose Predictioncited by 0
  3. 2026Representing 3D Faces with Learnable B-Spline Volumescited by 0