English

Back to Optimization: Diffusion-based Zero-Shot 3D Human Pose Estimation

Computer Vision and Pattern Recognition 2023-10-26 v3 Artificial Intelligence

Abstract

Learning-based methods have dominated the 3D human pose estimation (HPE) tasks with significantly better performance in most benchmarks than traditional optimization-based methods. Nonetheless, 3D HPE in the wild is still the biggest challenge for learning-based models, whether with 2D-3D lifting, image-to-3D, or diffusion-based methods, since the trained networks implicitly learn camera intrinsic parameters and domain-based 3D human pose distributions and estimate poses by statistical average. On the other hand, the optimization-based methods estimate results case-by-case, which can predict more diverse and sophisticated human poses in the wild. By combining the advantages of optimization-based and learning-based methods, we propose the \textbf{Ze}ro-shot \textbf{D}iffusion-based \textbf{O}ptimization (\textbf{ZeDO}) pipeline for 3D HPE to solve the problem of cross-domain and in-the-wild 3D HPE. Our multi-hypothesis \textit{\textbf{ZeDO}} achieves state-of-the-art (SOTA) performance on Human3.6M, with minMPJPE 51.451.4mm, without training with any 2D-3D or image-3D pairs. Moreover, our single-hypothesis \textit{\textbf{ZeDO}} achieves SOTA performance on 3DPW dataset with PA-MPJPE 40.340.3mm on cross-dataset evaluation, which even outperforms learning-based methods trained on 3DPW.

Keywords

Cite

@article{arxiv.2307.03833,
  title  = {Back to Optimization: Diffusion-based Zero-Shot 3D Human Pose Estimation},
  author = {Zhongyu Jiang and Zhuoran Zhou and Lei Li and Wenhao Chai and Cheng-Yen Yang and Jenq-Neng Hwang},
  journal= {arXiv preprint arXiv:2307.03833},
  year   = {2023}
}

Comments

WACV 2024

R2 v1 2026-06-28T11:24:54.314Z