mlsanityData to 14 Oct 2026

PARTFIELD: Learning 3D Feature Fields for Part Segmentation and Beyond

arXiv 2504.11451 ↗First posted 15 Apr 2025cs.CVcited by 95

Not on the map. Why?

Abstract (arXiv)

We propose PartField, a feedforward approach for learning part-based 3D features, which captures the general concept of parts and their hierarchy without relying on predefined templates or text-based names, and can be applied to open-world 3D shapes across various modalities. PartField requires only a 3D feedforward pass at inference time, significantly improving runtime and robustness compared to prior approaches. Our model is trained by distilling 2D and 3D part proposals from a mix of labeled datasets and image segmentations on large unsupervised datasets, via a contrastive learning formulation. It produces a continuous feature field which can be clustered to yield a hierarchical part decomposition. Comparisons show that PartField is up to 20% more accurate and often orders of magnitude faster than other recent class-agnostic part-segmentation methods. Beyond single-shape part decomposition, consistency in the learned field emerges across shapes, enabling tasks such as co-segmentation and correspondence, which we demonstrate in several applications of these general-purpose, hierarchical, and consistent 3D feature fields. Check our Webpage! https://research.nvidia.com/labs/toronto-ai/partfield-release/

Key facts. Cited by 95 papers in the map (data through 14 Oct 2026). Builds on 68 of its 74 references that are in the map. First posted on arXiv 15 Apr 2025.

Builds on

68 of 74 references.

References · 68

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 12,049
  3. 2023Segment Anythingcited by 8,234
  4. 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
  5. 20233D Gaussian Splatting for Real-Time Radiance Field Renderingcited by 6,226
  6. 2017PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Spacecited by 4,401
  7. 2024SAM 2: Segment Anything in Images and Videoscited by 2,863
  8. 2018Dynamic Graph CNN for Learning on Point Cloudscited by 2,223
  9. 2022Objaverse: A Universe of Annotated 3D Objectscited by 1,540
  10. 2019KPConv: Flexible and Deformable Convolution for Point Cloudscited by 1,205
  11. 2021Grounded Language-Image Pre-trainingcited by 1,156
  12. 2024Structured 3D Latents for Scalable and Versatile 3D Generationcited by 770
38 more
  1. 2020SAPIEN: A SimulAted Part-based Interactive ENvironmentcited by 595
  2. 2022OpenScene: 3D Scene Understanding with Open Vocabulariescited by 559
  3. 2023LERF: Language Embedded Radiance Fieldscited by 543
  4. 2018PartNet: A Large-scale Benchmark for Fine-grained and Hierarchical Part-level 3D Object Understandingcited by 526
  5. 2021PointCLIP: Point Cloud Understanding by CLIPcited by 463
  6. 2022PointNeXt: Revisiting PointNet++ with Improved Training and Scaling Strategiescited by 436
  7. 2023LangSplat: 3D Language Gaussian Splattingcited by 415
  8. 2023Gaussian Grouping: Segment and Edit Anything in 3D Scenescited by 393
  9. 2023Feature 3DGS: Supercharging 3D Gaussian Splatting to Enable Distilled Feature Fieldscited by 364
  10. 2019Point-Voxel CNN for Efficient 3D Deep Learningcited by 343
  11. 2018ABC: A Big CAD Model Dataset For Geometric Deep Learningcited by 341
  12. 2020PointGroup: Dual-Set Point Grouping for 3D Instance Segmentationcited by 321
  13. 2023OpenMask3D: Open-Vocabulary 3D Instance Segmentationcited by 289
  14. 2022SoftGroup for 3D Instance Segmentation on Point Cloudscited by 200
  15. 2022PointCLIP V2: Prompting CLIP and GPT for Powerful 3D Open-world Learningcited by 187
  16. 2023SAM3D: Segment Anything in 3D Scenescited by 157
  17. 2023Open3DIS: Open-Vocabulary 3D Instance Segmentation with 2D Mask Guidancecited by 124
  18. 2022PartSLIP: Low-Shot Part Segmentation for 3D Point Clouds via Pretrained Image-Language Modelscited by 118
  19. 2024GARField: Group Anything with Radiance Fieldscited by 109
  20. 2024SAMPart3D: Segment Any Part in 3D Objectscited by 97
  21. 2023SAI3D: Segment Any Instance in 3D Scenescited by 90
  22. 2023OpenIns3D: Snap and Lookup for 3D Open-vocabulary Instance Segmentationcited by 82
  23. 2023OVIR-3D: Open-Vocabulary 3D Instance Retrieval Without Training on 3D Datacited by 81
  24. 2023OmniSeg3D: Omniversal 3D Segmentation via Hierarchical Contrastive Learningcited by 75
  25. 2023Segment3D: Learning Fine-Grained Class-Agnostic 3D Segmentation without Manual Labelscited by 64
  26. 2023SATR: Zero-Shot Semantic Segmentation of 3D Shapescited by 62
  27. 2023Segment Anything in 3D with Radiance Fieldscited by 62
  28. 2024FlashSplat: 2D to 3D Gaussian Splatting Segmentation Solved Optimallycited by 58
  29. 2024Part123: Part-aware 3D Reconstruction from a Single-view Imagecited by 55
  30. 2024Point-SAM: Promptable 3D Segmentation Model for Point Cloudscited by 50
  31. 2024EgoLifter: Open-world 3D Segmentation for Egocentric Perceptioncited by 47
  32. 2019PartNet: A Recursive Part Decomposition Network for Fine-grained and Hierarchical Shape Segmentationcited by 44
  33. 2023SAMPro3D: Locating SAM Prompts in 3D for Zero-Shot Instance Segmentationcited by 44
  34. 2024EmbodiedSAM: Online Segment Any 3D Thing in Real Timecited by 40
  35. 2022Smooth Non-Rigid Shape Matching via Effective Dirichlet Energy Optimizationcited by 37
  36. 2023PartSLIP++: Enhancing Low-Shot 3D Part Segmentation via Multi-View Instance Segmentation and Maximum Likelihood Estimationcited by 37
  37. 2024Segment Any Meshcited by 37
  38. 2023SAM-guided Graph Cut for 3D Instance Segmentationcited by 35

50 most cited of 68 All 68 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 34; 2026: 61.

95 papers.

Sort:

By area:

  1. 2025Efficient Part-level 3D Object Generation via Dual Volume Packingcited by 48
  2. 2025Ultra3D: Efficient and High-Fidelity 3D Generation with Part Attentioncited by 44
  3. 2025X-Part: high fidelity and structure coherent shape decompositioncited by 44
  4. 2025Puppeteer: Rig and Animate Your 3D Modelscited by 43
  5. 2025PhysX-3D: Physical-Grounded 3D Asset Generationcited by 42
  6. 2025IGGT: Instance-Grounded Geometry Transformer for Semantic 3D Reconstructioncited by 28
  7. 2025PartNeXt: A Next-Generation Dataset for Fine-Grained and Hierarchical 3D Part Understandingcited by 26
  8. 2025Particulate: Feed-Forward 3D Object Articulationcited by 23
  9. 2025iTACO: Interactable Digital Twins of Articulated Objects from Casually Captured RGBD Videoscited by 19
  10. 2025WorldGen: From Text to Traversable and Interactive 3D Worldscited by 19
  11. 2025PartSAM: A Scalable Promptable Part Segmentation Model Trained on Native 3D Datacited by 16
  12. 2025Rectified Point Flow: Generic Point Cloud Pose Estimationcited by 14
38 more
  1. 2025GeoSAM2: Unleashing the Power of SAM2 for 3D Part Segmentationcited by 13
  2. 2026HY3D-Bench: Generation of 3D Assetscited by 11
  3. 2025PhysWorld: From Real Videos to World Models of Deformable Objects via Physics-Aware Demonstration Synthesiscited by 10
  4. 2026SegviGen: Repurposing 3D Generative Model for Part Segmentationcited by 10
  5. 2026Feedforward 3D Editing Learns from Semantic-Part Transformationcited by 10
  6. 2025PhysChoreo: Physics-Controllable Video Generation with Part-Aware Semantic Groundingcited by 9
  7. 2025UniPart: Part-Level 3D Generation with Unified 3D Geom-Seg Latentscited by 9
  8. 2026PhysForge: Generating Physics-Grounded 3D Assets for Interactive Virtual Worldcited by 9
  9. 2025MeshMosaic: Scaling Artist Mesh Generation via Local-to-Global Assemblycited by 8
  10. 2025Part-X-MLLM: Part-aware 3D Multimodal Large Language Modelcited by 7
  11. 2026GaussianFluent: Gaussian Simulation for Dynamic Scenes with Mixed Materialscited by 7
  12. 2025Affogato: Open-Vocabulary Affordance Grounding with Automated Data Generation at Scalecited by 6
  13. 2026PatchAlign3D: Local Feature Alignment for Dense 3D Shape Understandingcited by 6
  14. 2026Feed-Forward 3D Scene Modeling: A Problem-Driven Perspectivecited by 6
  15. 2025Residual Primitive Fitting of 3D Shapes with SuperFrustacited by 5
  16. 2026MonoArt: Progressive Structural Reasoning for Monocular Articulated 3D Reconstructioncited by 5
  17. 2026SIMART: Decomposing Monolithic Meshes into Sim-ready Articulated Assets via MLLMcited by 5
  18. 2025PartDiffuser: Part-wise 3D Mesh Generation via Discrete Diffusioncited by 4
  19. 2025TEXTRIX: Latent Attribute Grid for Native Texture Generation and Beyondcited by 4
  20. 2026DextER: Language-driven Dexterous Grasp Generation with Embodied Reasoningcited by 4
  21. 2026ArtHOI: Taming Foundation Models for Monocular 4D Reconstruction of Hand-Articulated-Object Interactionscited by 4
  22. 2026MeshTailor: Cutting Seams via Generative Mesh Traversalcited by 4
  23. 2026Instruct-Particulate: Scaling Feed-Forward 3D Object Articulation with Kinematic Controlcited by 4
  24. 2026Procedura: Agentic 3D Modeling with Procedural Controlcited by 4
  25. 2025Hi3DEval: Advancing 3D Generation Evaluation with Hierarchical Validitycited by 3
  26. 2025Name That Part: 3D Part Segmentation and Namingcited by 3
  27. 2026Vinedresser3D: Agentic Text-guided 3D Editingcited by 3
  28. 2026Interact3D: Compositional 3D Generation of Interactive Objectscited by 3
  29. 2026Hierarchical Point-Patch Fusion with Adaptive Patch Codebook for 3D Shape Anomaly Detectioncited by 3
  30. 2026Strips as Tokens: Artist Mesh Generation with Native UV Segmentationcited by 3
  31. 2025GuideFlow3D: Optimization-Guided Rectified Flow For Appearance Transfercited by 2
  32. 2025ZeroDexGrasp: Zero-Shot Task-Oriented Dexterous Grasp Synthesis with Prompt-Based Multi-Stage Semantic Reasoningcited by 2
  33. 2025UniArt: Unified 3D Representation for Generating 3D Articulated Objects with Open-Set Articulationcited by 2
  34. 2026Learning Convex Decomposition via Feature Fieldscited by 2
  35. 2026MotionAnymesh: Physics-Grounded Articulation for Simulation-Ready Digital Twinscited by 2
  36. 2026TriFlow: Generating Artist-Like 3D Mesh Topology via Nearest-Vertex Vector Fieldscited by 2
  37. 2026UniPhysGen: Unified Physical Grounding for Simulation-Ready 3D Assetscited by 2
  38. 2025Unsupervised Representation Learning for 3D Mesh Parameterization with Semantic and Visibility Objectivescited by 1

50 most cited of 95 All 95 (JSON)