mlsanityData to 14 Oct 2026

Fast SAM 3D Body: Accelerating SAM 3D Body for Real-Time Full-Body Human Mesh Recovery

arXiv 2603.15603 ↗First posted 16 Mar 2026cs.CVcited by 4

Not on the map. Why?

Abstract (arXiv)

SAM 3D Body (3DB) achieves state-of-the-art accuracy in monocular 3D human mesh recovery, yet its inference latency of several seconds per image precludes real-time application. We present Fast SAM 3D Body, a training-free acceleration framework that reformulates the 3DB inference pathway to achieve interactive rates. By decoupling serial spatial dependencies and applying architecture-aware pruning, we enable parallelized multi-crop feature extraction and streamlined transformer decoding. Moreover, to extract the joint-level kinematics (SMPL) compatible with existing humanoid control and policy learning frameworks, we replace the iterative mesh fitting with a direct feedforward mapping, accelerating this specific conversion by over 10,000x. Overall, our framework delivers up to a 10.9x end-to-end speedup while maintaining on-par reconstruction fidelity, even surpassing 3DB on benchmarks such as LSPET. We demonstrate its utility by deploying Fast SAM 3D Body in a vision-only teleoperation system that-unlike methods reliant on wearable IMUs-enables real-time humanoid control and the direct collection of manipulation policies from a single RGB stream.

Key facts. Cited by 4 papers in the map (data through 14 Oct 2026). Builds on 40 of its 47 references that are in the map. First posted on arXiv 16 Mar 2026.

Builds on

40 of 47 references.

References · 40

  1. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  2. 2019Expressive Body Capture: 3D Hands, Face, and Body from a Single Imagecited by 1,612
  3. 2025DINOv3cited by 1,414
  4. 2019AMASS: Archive of Motion Capture as Surface Shapescited by 1,236
  5. 2022Token Merging: Your ViT But Fastercited by 1,021
  6. 2017End-to-end Recovery of Human Shape and Posecited by 885
  7. 2022Exploring Plain Vision Transformer Backbones for Object Detectioncited by 724
  8. 2023Humans in 4D: Reconstructing and Tracking Humans with Transformerscited by 350
  9. 2023Reconstructing Hands in 3D with Transformerscited by 306
  10. 2019FreiHAND: A Dataset for Markerless Capture of Hand Pose and Shape from Single RGB Imagescited by 271
  11. 2025MoGe-2: Accurate Monocular Geometry with Metric Scale and Sharp Detailscited by 261
  12. 2022CLIFF: Carrying Location Information in Full Frames into Human Pose and Shape Estimationcited by 238
28 more
  1. 2025EgoDex: Learning Dexterous Manipulation from Large-Scale Egocentric Videocited by 211
  2. 2023WHAM: Reconstructing World-grounded Humans with Accurate 3D Motioncited by 206
  3. 2022Capturing and Inferring Dense Full-Body Human-Scene Contactcited by 180
  4. 2023RTMPose: Real-Time Multi-Person Pose Estimation based on MMPosecited by 175
  5. 2020Monocular Expressive Body Regression through Body-Driven Attentioncited by 155
  6. 2023SMPLer-X: Scaling Up Expressive Human Pose and Shape Estimationcited by 155
  7. 2024WiLoR: End-to-end 3D Hand Localization and Reconstruction in-the-wildcited by 124
  8. 2024TRAM: Global Trajectory and Motion of 3D Humans from in-the-wild Videoscited by 106
  9. 2024Multi-HMR: Multi-Person Whole-Body Human Mesh Recovery in a Single Shotcited by 101
  10. 2025SONIC: Supersizing Motion Tracking for Natural Humanoid Whole-Body Controlcited by 101
  11. 2023EMDB: The Electromagnetic Database of Global 3D Human Pose and Shape in the Wildcited by 97
  12. 2024TokenHMR: Advancing Human Mesh Recovery with a Tokenized Pose Representationcited by 95
  13. 2022Cross-Attention of Disentangled Modalities for 3D Human Mesh Recovery with Transformerscited by 81
  14. 2025PromptHMR: Promptable Human Mesh Recoverycited by 76
  15. 2026SAM 3D Body: Robust Full-Body Human Mesh Recoverycited by 69
  16. 2024CameraHMR: Aligning People with Perspectivecited by 65
  17. 2024Neural Localizer Fields for Continuous 3D Human Pose and Shape Estimationcited by 59
  18. 2025GENMO: A GENeralist Model for Human MOtioncited by 49
  19. 2022TORE: Token Reduction for Efficient Human Mesh Recovery with Transformercited by 41
  20. 2020Inference Stage Optimization for Cross-scenario 3D Human Pose Estimationcited by 35
  21. 2022Expediting Large-Scale Vision Transformer for Dense Prediction without Fine-tuningcited by 33
  22. 2023POTTER: Pooling Attention Transformer for Efficient Human Mesh Recoverycited by 31
  23. 2024Harmony4D: A Video Dataset for In-The-Wild Close Human Interactionscited by 31
  24. 2026Psi0: An Open Foundation Model Towards Universal Humanoid Loco-Manipulationcited by 27
  25. 2025Humanoid Everyday: A Comprehensive Robotic Dataset for Open-World Humanoid Manipulationcited by 25
  26. 2022FeatER: An Efficient Network for Human Reconstruction via Feature Map-Based TransformERcited by 9
  27. 2021Fast and Flexible Human Pose Estimation with HyperPosecited by 5
  28. 2025FastHMR: Accelerating Human Mesh Recovery via Token and Layer Merging with Diffusion Decodingcited by 4

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 4.

4 papers.

  1. 2026GenPHRI: Agentic Generative Simulation for Physical Human-Robot Interactioncited by 0
  2. 2026Let RGB Be the Language of Visioncited by 0
  3. 2026Semantic Audio-driven Understanding for Dynamic Humanoid Whole Body Controlcited by 0
  4. 2026Reliability-Gated Fusion of Consumer Head and Foot IMUs for Lower-Body 3D Posecited by 0