The Third Workshop on
3D Human Understanding:
Towards Human-centric World Models
A 2026 Perspective on 3D Human Perception, World Models, and Robotics
1.50PM - 5.30PM, September 8
ECCV 2026
Malmo, Sweden
Embedded Files
Speakers
Meta
Seoul National University
University of Bristol
ETH Zürich
Nvidia
Overview
Overview
The Third HUMANS spans core problems in human modeling, including motion capture, 3D reconstruction from images and videos, and human–scene–object interaction, while situating these advances within the broader goal of developing human-centric world models capable of long-horizon behavior and interaction reasoning. As computer vision increasingly focuses on large-scale video modeling and interactive simulation, these capabilities become essential for representing people within dynamic environments. A central theme of the workshop is bridging recognition and generation, from reconstructing pose and shape to modeling motion dynamics and controllable synthesis. Progress in physics-aware modeling, multimodal alignment, and scalable data generation supports more coherent and predictive representations of humans in context, enabling applications such as XR, digital avatars, and interactive simulation.
Social Behaviors
Social Behaviors
Egocentric Vision
Egocentric Vision
Crowd Behaviors
Crowd Behaviors
Embodied Intelligence
Embodied Intelligence
Automonous Driving
Automonous Driving
Human Robot Collaboration
Human Robot Collaboration
Tentative Schedule
13:50 - 14:00 Welcome and Opening Remark
14:00 - 14:30 Talk #1: Shunsuke Saito
Title Coming Soon!
14:30 - 15:00 Talk #2: Hanbyul Joo
Title Coming Soon!
15:00 - 15:30 Talk #3: Dima Damen
Modelling in an Ego-Sensed World
15:30 - 16:00 Talk #4: Siyu Tang
Title Coming Soon!
16:00 - 16:30 Talk #5: Umar Iqbal
Title Coming Soon!
16:30 - 17:30 Panel discussion: Georgios Pavlakos + All Speakers
Organization Team
Organizers
Senior Advisory Board
Page updated
Google Sites
Report abuse