AI Paper Atlas

Research map / Benchmarks / Shapes3D

Shapes3D leaderboard

10 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on Shapes3D. Best DCI: 0.973 (Ours (flow matching + orth), NeurIPS 2026).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

Shapes3D · DCI

10 papers, each with its best reported number for its own method; higher is better.

#DCI ↑MethodPaperTopic
10.973Ours (flow matching + orth)Disentangled Representation Learning via Flow Matching
NeurIPS 2026 · Table 1
Molecules and proteins
20.969EncDiffDiffusion Model with Cross Attention as an Inductive Bias for Disentanglement
NeurIPS 2024 · Table 1
Autoregressive and latent generation
30.917FDAEFactorized Diffusion Autoencoder for Unsupervised Disentangled Representation Learning
AAAI 2024 · Table 2
Autoregressive and latent generation
40.884VCT (Ours)Visual Concepts Tokenization
NeurIPS 2022 · Table 1
Human and embodied modeling
50.844DisCo (VAE)
VAE
Learning Disentangled Representation by Exploiting Pretrained Generative Models: A Contrastive Learning View
ICLR 2022 · Table 1
Autoregressive and latent generation
60.837OursDisentangled Representation Learning via Modular Compositional Bias
NeurIPS 2025 · Appendix comparison table
Recognition and long-tail
70.78DAVADAVA: Disentangling Adversarial Variational Autoencoder
ICLR 2023 · Table Results Shapes3D
Variational and generative sampling
80.731CL-Dis (Ours)
Diff-AE
Closed-Loop Unsupervised Representation Disentanglement with $\\beta$-VAE Distillation and Diffusion Probabilistic Feedback
ECCV 2024 · Table 3
Autoregressive and latent generation
90.723DisDiff-VQDisDiff: Unsupervised Disentanglement of Diffusion Probabilistic Models
NeurIPS 2023 · Table 1
Fast diffusion sampling
100.72Groupified β-TCVAE
β-TCVAE
Towards Building A Group-based Unsupervised Representation Disentanglement Framework
ICLR 2022 · Table 1
Recognition and long-tail

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.