Fumin Shen

About

Fumin Shen is a leading researcher at the intersection of computer vision and human-robot interaction (HRI), with a core focus on human action recognition and embodied AI. His most impactful contributions address the critical challenge of enabling robots to understand human actions from arbitrary viewpoints—a fundamental requirement for natural, fluent HRI. Shen pioneered the creation of large-scale, varying-view RGB-D action datasets, including a landmark database with 77 citations, which has become a vital resource for the field. His widely cited survey on human action analysis in HRI (89 citations) provides a comprehensive roadmap for the community. Beyond action recognition, Shen has advanced cross-domain facial expression recognition through novel feature fusion networks (84 citations) and tackled the complex problem of Embodied Question Answering (EQA) with his robust learning framework, "Robust-EQA" (17 citations). By systematically addressing the limitations of single- and multi-view recognition, Shen’s work has laid the groundwork for robots that can perceive and interact with humans in unconstrained, real-world environments, making him a pivotal figure in the evolution of socially aware robotics.

Research Focus

Key Achievements

Top Papers

1. 1
   A Survey of Human Action Analysis in HRI Applications

   89 citations · 2019
2. 2
   Cross-domain facial expression recognition via an intra-category common feature and inter-category Distinction feature fusion network

   84 citations · 2018
3. 3
   A Large-scale RGB-D Database for Arbitrary-view Human Action Recognition

   77 citations · 2018
4. 4
   Arbitrary-View Human Action Recognition: A Varying-View RGB-D Action Dataset

   39 citations · 2020
5. 5
   A Large-scale Varying-view RGB-D Action Dataset for Arbitrary-view Human Action Recognition

   18 citations · 2019
6. 6
   Robust-EQA: Robust Learning for Embodied Question Answering With Noisy Labels

   17 citations · 2023

Key Collaborators

Contact & Links