# PriorFormer: A Transformer for Real-time Monocular 3D Human Pose Estimation with Versatile Geometric Priors

arXiv: https://arxiv.org/abs/2508.18238 · HTML: https://mlsanity.com/paper/2508.18238/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 21 of its 30 references that are in the map.
- First posted on arXiv 21 Aug 2025.

## Builds on (most cited first, 21 of 21)

- [Deep High-Resolution Representation Learning for Human Pose Estimation](https://mlsanity.com/paper/1902.09212/) (2019)
- [AMASS: Archive of Motion Capture as Surface Shapes](https://mlsanity.com/paper/1904.03278/) (2019)
- [Coarse-to-Fine Volumetric Prediction for Single-Image 3D Human Pose](https://mlsanity.com/paper/1611.07828/) (2016)
- [RTMPose: Real-Time Multi-Person Pose Estimation based on MMPose](https://mlsanity.com/paper/2303.07399/) (2023)
- [MHFormer: Multi-Hypothesis Transformer for 3D Human Pose Estimation](https://mlsanity.com/paper/2111.12707/) (2021)
- [PoseFormerV2: Exploring Frequency Domain for Efficient and Robust 3D Human Pose Estimation](https://mlsanity.com/paper/2303.17472/) (2023)
- [HOVER: Versatile Neural Whole-Body Controller for Humanoid Robots](https://mlsanity.com/paper/2410.21229/) (2024)
- [ViTPose++: Vision Transformer for Generic Body Pose Estimation](https://mlsanity.com/paper/2212.04246/) (2022)
- [CameraHMR: Aligning People with Perspective](https://mlsanity.com/paper/2411.08128/) (2024)
- [MoVi: A Large Multipurpose Motion and Video Dataset](https://mlsanity.com/paper/2003.01888/) (2020)
- [Uplift and Upsample: Efficient 3D Human Pose Estimation with Uplifting Transformers](https://mlsanity.com/paper/2210.06110/) (2022)
- [Self-Supervised Camera Self-Calibration from Video](https://mlsanity.com/paper/2112.03325/) (2021)
- [HumanNeRF-SE: A Simple yet Effective Approach to Animate HumanNeRF with Diverse Poses](https://mlsanity.com/paper/2312.02232/) (2023)
- [Lifting Motion to the 3D World via 2D Diffusion](https://mlsanity.com/paper/2411.18808/) (2024)
- [Jointformer: Single-Frame Lifting Transformer with Error Prediction and Refinement for 3D Human Pose Estimation](https://mlsanity.com/paper/2208.03704/) (2022)
- [OpenCapBench: A Benchmark to Bridge Pose Estimation and Biomechanics](https://mlsanity.com/paper/2406.09788/) (2024)
- [Lifting 2D Human Pose to 3D with Domain Adapted 3D Body Concept](https://mlsanity.com/paper/2111.11969/) (2021)
- [BLADE: Single-view Body Mesh Learning through Accurate Depth Estimation](https://mlsanity.com/paper/2412.08640/) (2024)
- [Estimation of 3D Human Pose Using Prior Knowledge](https://mlsanity.com/paper/2105.03807/) (2021)
- [3D WholeBody Pose Estimation based on Semantic Graph Attention Network and Distance Information](https://mlsanity.com/paper/2406.01196/) (2024)
- [AirNeRF: 3D Reconstruction of Human with Drone and NeRF for Future Communication Systems](https://mlsanity.com/paper/2407.10865/) (2024)

## Built on by (most cited first, 1 of 1)

- [SIMSPINE: A Biomechanics-Aware Simulation Framework for 3D Spine Motion Annotation and Benchmarking](https://mlsanity.com/paper/2602.20792/) (2026)

## Abstract (arXiv)

This paper proposes a new lightweight Transformer-based lifter that maps short sequences of human 2D joint positions to 3D poses using a single camera. The proposed model takes as input geometric priors including segment lengths and camera intrinsics and is designed to operate in both calibrated and uncalibrated settings. To this end, a masking mechanism enables the model to ignore missing priors during training and inference. This yields a single versatile network that can adapt to different deployment scenarios, from fully calibrated lab environments to in-the-wild monocular videos without calibration. The model was trained using 3D keypoints from AMASS dataset with corresponding 2D synthetic data generated by sampling random camera poses and intrinsics. It was then compared to an expert model trained, only on complete priors, and the validation was done by conducting an ablation study. Results show that both, camera and segment length priors, improve performance and that the versatile model outperforms the expert, even when all priors are available, and maintains high accuracy when priors are missing. Overall the average 3D joint center positions estimation accuracy was as low as 36mm improving state of the art by half a centimeter and at a much lower computational cost. Indeed, the proposed model runs in 380μs on GPU and 1800μs on CPU, making it suitable for deployment on embedded platforms and low-power devices.