Congratulations to the international team for winning the championship of HANDS Challenge at European Conference on Computer Vision (ECCV) 2022 at “Task 2: Semi/Self-supervised Two-hands 3D Pose Estimation during Hand-object and Hand-hand interactions”. The task aims to address two-hands 3D pose estimation problems with multi-view captured videos of human-object interaction, which is critical for building AR/VR system.

The absence of 3D pose annotation is the main difficult part for training as 3D results can only be reconstructed from the multi-view 2D counterparts with a semi/self-supervised learning strategy. To address this, the team proposed a closed-form solution from 2D to 3D, without requiring any training procedure. It has achieved the leading performance with high running efficiency and would not suffer from any result jittering due to training’s randomness. The team believes the method could essentially facilitate the applications towards metaverse to advance Web 3.0. 

Team Members:

	Mr. Wu Cunlin School of Artificial Intelligence and Automation Huazhong University of Science and Technology, China		Dr. Xiao Yang School of Artificial Intelligence and Automation Huazhong University of Science and Technology, China
	Mr. Jiang Changlong School of Artificial Intelligence and Automation Huazhong University of Science and Technology, China		Mr. Zheng Jinghong School of Artificial Intelligence and Automation Huazhong University of Science and Technology, China
	Dr. Cao Zhiguo School of Artificial Intelligence and Automation Huazhong University of Science and Technology, China		Dr. Fang Zhiwen Southern Medical University, China
	Dr. Joey Zhou Tianyi  A*STAR Centre for Frontier AI Research (CFAR)		Dr. Yuan Junsong Department of Computer Science and Engineering (CSE) State University of New York at Buffalo, USA