Riemannian Graph Convolutional Network for Skeleton-Based Two-Person Interaction Recognition

Wang, Rui; Bi, Zihao; Hu, Chen; Song, Xiaoning; Wu, Xiao-Jun; Sebe, Nicu; Chen, Ziheng

doi:10.24963/ijcai.2026/196

Riemannian Graph Convolutional Network for Skeleton-Based Two-Person Interaction Recognition

Rui Wang, Zihao Bi, Chen Hu, Xiaoning Song, Xiao-Jun Wu, Nicu Sebe, Ziheng Chen

Download BibTeX Cite

Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence. Main Track. Pages 1758-1766.

https://doi.org/10.24963/ijcai.2026/196

Keywords

Machine Learning: Geometric learning
Machine Learning: Deep learning architectures
Computer Vision: Action and behavior recognition
Computer Vision: Machine learning for vision

Abstract

In the field of skeleton-based human action recognition, Graph Convolutional Networks (GCNs) have become a dominant framework. However, existing GCN-based approaches often treat the sequences of two-person interaction as separate entities, ignoring the inherent semantic dependencies and spatial correlations between interacting subjects. Furthermore, high-order skeleton representations naturally exhibit non-Euclidean structures, where Euclidean deep learning models are inherently limited in explicitly capturing and preserving such geometric information. As a countermeasure, we propose a Riemannian Graph Convolutional Network (RGCN) that operates on the Symmetric Positive Definite (SPD) manifolds. Specifically, we model high-order skeletal statistics via Gaussian embedding and propose a Riemannian network to capture inter-subject interactions and global correlations. The proposed RGCN is instantiated under three SPD geometries, and its effectiveness is validated through extensive experiments on three interaction benchmarks. Extensive experimental results show that RGCN provides a competitive and geometrically grounded alternative for skeleton-based interaction recognition.

Rui Wang, Zihao Bi, Chen Hu, Xiaoning Song, Xiao-Jun Wu, Nicu Sebe, and Ziheng Chen. 2026. Riemannian Graph Convolutional Network for Skeleton-Based Two-Person Interaction Recognition. In Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence, IJCAI-26. IJCAI, 1758–1766. https://doi.org/10.24963/ijcai.2026/196

@inproceedings{ijcai2026p196,
  title     = {Riemannian Graph Convolutional Network for Skeleton-Based Two-Person Interaction Recognition},
  author    = {Wang, Rui and Bi, Zihao and Hu, Chen and Song, Xiaoning and Wu, Xiao-Jun and Sebe, Nicu and Chen, Ziheng},
  booktitle = {Proceedings of the Thirty-Fifth International Joint Conference on
               Artificial Intelligence, {IJCAI-26}},
  publisher = {International Joint Conferences on Artificial Intelligence Organization},
  editor    = {Diego Calvanese},
  pages     = {1758--1766},
  year      = {2026},
  month     = {8},
  note      = {Main Track},
  doi       = {10.24963/ijcai.2026/196},
  url       = {https://doi.org/10.24963/ijcai.2026/196},
}

TY  - CPAPER
TI  - Riemannian Graph Convolutional Network for Skeleton-Based Two-Person Interaction Recognition
AU  - Wang, Rui
AU  - Bi, Zihao
AU  - Hu, Chen
AU  - Song, Xiaoning
AU  - Wu, Xiao-Jun
AU  - Sebe, Nicu
AU  - Chen, Ziheng
T2  - Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence, IJCAI-26
PB  - International Joint Conferences on Artificial Intelligence Organization
PY  - 2026
SP  - 1758
EP  - 1766
DO  - 10.24963/ijcai.2026/196
UR  - https://doi.org/10.24963/ijcai.2026/196
ER  -