mlsanityData to 7 Oct 2026

KeyPointDiffuser: Unsupervised 3D Keypoint Learning via Latent Diffusion Models

arXiv 2512.03450 ↗First posted 3 Dec 2025cs.CV cs.LGcited by 0

On the map: Diffusion models → 3D, point cloud and scene generation

Abstract (arXiv)

Understanding and representing the structure of 3D objects in an unsupervised manner remains a core challenge in computer vision and graphics. Most existing unsupervised keypoint methods are not designed for unconditional generative settings, restricting their use in modern 3D generative pipelines; our formulation explicitly bridges this gap. We present an unsupervised framework for learning spatially structured 3D keypoints from point cloud data. These keypoints serve as a compact and interpretable representation that conditions an Elucidated Diffusion Model (EDM) to reconstruct the full shape. The learned keypoints exhibit repeatable spatial structure across object instances and support smooth interpolation in keypoint space, indicating that they capture geometric variation. Our method achieves strong performance across diverse object categories, yielding a 6 percentage-point improvement in keypoint consistency compared to prior approaches.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in 3D, point cloud and scene generation, one of 14 directions in Diffusion models. Builds on 30 of its 41 references that are in the map. First posted on arXiv 3 Dec 2025.

Builds on

30 of 41 references.

Closest ancestors · 3

Same direction, newest first

  1. 2023PolyDiff: Generating 3D Polygonal Meshes with Diffusion Models18 Dec
  2. 2022LION: Latent Point Diffusion Models for 3D Shape Generation12 Oct
  3. 2021Diffusion Probabilistic Models for 3D Point Cloud Generation2 Mar

Other references · 27

  1. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,694
  2. 2022Elucidating the Design Space of Diffusion-Based Generative Modelscited by 2,994
  3. 2015ShapeNet: An Information-Rich 3D Model Repositorycited by 2,880
  4. 2017FiLM: Visual Reasoning with a General Conditioning Layercited by 2,286
  5. 2021Variational Diffusion Modelscited by 1,145
  6. 2020Object-Centric Learning with Slot Attentioncited by 865
  7. 2016Learning a Probabilistic Latent Space of Object Shapes via 3D Generative-Adversarial Modelingcited by 558
  8. 2025REPA-E: Unlocking VAE for End-to-End Tuning with Latent Diffusion Transformerscited by 212
  9. 2021EgoBody: Human Body Shape and Motion of Interacting People from Head-Mounted Devicescited by 136
  10. 2023Neural Kernel Surface Reconstructioncited by 102
  11. 2021Learnable Fourier Features for Multi-Dimensional Spatial Positional Encodingcited by 93
  12. 2018Discovery of Latent 3D Keypoints via End-to-end Geometric Reasoningcited by 79
15 more
  1. 2018Unsupervised Discovery of Object Landmarks as Structural Representationscited by 73
  2. 2017Unsupervised learning of object landmarks by factorized spatial embeddingscited by 64
  3. 2019C3DPO: Canonical 3D Pose Networks for Non-Rigid Structure From Motioncited by 60
  4. 2018Deforming Autoencoders: Unsupervised Disentangling of Shape and Appearancecited by 58
  5. 2020KeypointNet: A Large-scale 3D Keypoint Dataset Aggregated from Numerous Human Annotationscited by 50
  6. 2021KeypointDeformer: Unsupervised 3D Keypoint Discovery for Shape Controlcited by 50
  7. 2021Skeleton Merger: an Unsupervised Aligned Keypoint Detectorcited by 27
  8. 2019Lifting AutoEncoders: Unsupervised Learning of a Fully-Disentangled 3D Morphable Model using Deep Non-Rigid Structure from Motioncited by 23
  9. 2020Unsupervised Learning of Category-Specific Symmetric 3D Keypoints from Point Setscited by 23
  10. 2020UKPGAN: A General Self-Supervised Keypoint Detectorcited by 15
  11. 2023SC3K: Self-supervised and Coherent 3D Keypoints Estimation from Rotated, Noisy, and Decimated Point Cloud Datacited by 9
  12. 2020Unsupervised Part Discovery by Unsupervised Disentanglementcited by 5
  13. 2024Key-Grid: Unsupervised 3D Keypoints Detection using Grid Heatmap Featurescited by 4
  14. 20233D Keypoint Estimation Using Implicit Representation Learningcited by 1
  15. 2024SelfGeo: Self-supervised and Geodesic-consistent Estimation of Keypoints on Deformable Shapescited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Unsupervised Monocular 3D Keypoint Discovery from Multi-View Diffusion Priors
  2. 2025Guided and Unguided Conditional Diffusion Mechanisms for Structured and Semantically-Aware 3D Point Cloud Generation
  3. 2023Unsupervised Keypoints from Pretrained Diffusion Models
  4. 2026Points-to-3D: Structure-Aware 3D Generation with Point Cloud Priors
  5. 2024PointInfinity: Resolution-Invariant Point Diffusion Models
  6. 2025StrucADT: Generating Structure-controlled 3D Point Clouds with Adjacency Diffusion Transformer
  7. 2024ZeroKey: Point-Level Reasoning and Zero-Shot 3D Keypoint Detection from Large Language Models
  8. 2024PCDreamer: Point Cloud Completion Through Multi-view Diffusion Priors