Global 3D Human Poses (G3P) Workshop

Summary

Over the past years, human pose estimation research has achieved remarkable progress—primarily focusing on capturing local, camera-centric body poses. Yet real-world applications demand a deeper understanding: one that integrates global spatial context and the full trajectories of human movement. In environments as diverse as robotics, sports analytics, virtual reality, and autonomous systems, the ability to accurately track and interpret human motion on a global scale is becoming increasingly essential.

Despite significant advances in deep learning and local pose estimation, methods that consider the complete, global context remain underexplored. This gap means that many state-of-the-art solutions, while impressive in controlled settings, fall short when applied to complex, dynamic scenarios where understanding motion in world coordinates is critical.

The Global 3D Human Poses (G3P) workshop is designed to bridge this gap by focusing on innovative techniques that incorporate trajectory data into pose estimation. By fostering collaboration among researchers and practitioners, the workshop will delve into new methodologies, address emerging challenges, and discuss the transformative potential of global pose estimation. Ultimately, the insights and innovations presented here are poised to push the boundaries of computer vision and pave the way for more robust, real-world applications in interactive systems and beyond.

Speakers

Prof. Michael J. Black

(MPI Perceiving Systems)

Prof. Hanbyul Joo

(Seoul National University)

Prof. Angjoo Kanazawa

(UC Berkeley)

Prof. Kris Kitani

(Carnegie Mellon University)

Schedule

Global Pose Challenge

Overview

We also host the competition from the FIFA Skeletal Tracking Innovation Programme. It features a public training split along with internally prepared validation and test splits. The challenge aims to advance the state-of-the-art in global pose estimation by evaluating submissions on a realistic and practical setting.

Results and Solutions

The challenge has now concluded. We thank all teams for their participation and contributions. You can find the top teams’ solution write-ups below:

Further Details

For more information about the challenge, please refer to the FIFA's Challenge Page for reference.

Organizers

Tianjian Jiang

ETH Zürich

Dr. Manuel Kaufmann

ETH Zürich

Prof. Jie Song

ETH Zürich & HKUST

Soyong Shin

Carnegie Mellon University

Jiye Lee

Seoul National University

Dr. Ye Yuan

NVIDIA Research

Chen Guo

ETH Zürich

Contact

For further inquiries, please email us at .