Yizhou Xu, WASP Industrial PhD student at KTH and Tracab, has received an outstanding paper award at the ECCV 2026 workshop “Human Motion-Informed World Models & Socially Intelligent Action.” The award recognizes outstanding papers presented at the workshop for their research contributions to advancing human motion understanding, world modeling, and socially intelligent AI systems.
The paper presents a method for recognizing soccer players’ actions from their 3D skeletal motion. It uses a context-aware approach to adapt pre-trained graph neural networks (GNNs) for this task.
“By leveraging pre-trained motion representations and incorporating game-specific contextual information such as ball trajectory and players’ positions on the pitch, our method improves action recognition in complex and dynamic soccer scenarios,” says Yizhou Xu.
For Xu, the award also provides encouragement to continue his research on human motion understanding in soccer.
“It means a lot to me to have our work recognized. I have really enjoyed working on human motion understanding in soccer, and receiving this award is encouraging. It gives me greater confidence that we can make meaningful changes in this area and motivates me to keep exploring how we can build AI systems that better understand soccer from tracking data.”
Workshop: https://human-motion-world-model.github.io/
Paper title: “Context-Aware Adaptation of Pre-trained GNNs for Skeleton-Based Soccer Action Understanding”. Link to paper
Authors: Yizhou Xu, Lars Bretzner, Tiesheng Wang, Atsuto Maki.
Published: October 22nd, 2026
[addtoany]