Download
SAM 3D Body is a promptable model for single-image full-body 3D human mesh recovery, designed to estimate detailed human pose and shape from just one RGB image. It reconstructs the full body, including feet and hands, using the Momentum Human Rig (MHR), a parametric mesh representation that decouples skeletal structure from surface shape for more accurate and interpretable results. The model is trained to be robust in diverse, in-the-wild conditions, so it handles varied clothing, viewpoints, and backgrounds while maintaining strong accuracy across multiple human-pose benchmarks. The repository provides Python code to run inference, utilities to download checkpoints from Hugging Face, and demo scripts that turn images into 3D meshes and visualizations. There are Jupyter notebooks that walk you through setting up the model, running it on example images, and visualizing outputs in 3D, making it approachable even if you are not a 3D expert.
Features
- Promptable full-body 3D human mesh recovery from a single RGB image
- Momentum Human Rig representation that separates skeleton and surface shape for better accuracy and interpretation
- Robust performance across in-the-wild images and multiple human-pose benchmarks
- Ready-to-use Python API, demo script, and estimator class for running inference on your own images
- Example Jupyter notebooks for visualization, benchmarking, and combining with SAM 3D Objects in a shared 3D frame
- Publicly released checkpoints and dataset under the SAM License for research and experimentation
Project Samples
