Heng Tao Shen

About

Heng Tao Shen is a prominent researcher specializing in human action recognition, human-robot interaction (HRI), and multimodal machine learning. His work has made substantial contributions to enabling robots and intelligent systems to understand and predict human behavior across diverse and challenging conditions. Shen is perhaps best known for advancing **arbitrary-view human action recognition**, a critical capability for real-world HRI applications. Recognizing the limitations of single- and multi-view approaches, he spearheaded the creation of large-scale RGB-D datasets designed specifically for arbitrary-view scenarios, work that has accumulated over 130 combined citations and become a foundational resource for the research community. His 2019 survey on human action analysis in HRI applications (89 citations) reflects his broad command of the field and serves as an essential reference for newcomers and experts alike. Beyond action recognition, Shen has contributed to facial expression recognition through innovative feature fusion networks, early action recognition via one-shot learning, and view-invariant recognition through attention transfer mechanisms. More recently, his research has expanded into embodied question answering and multimodal sentiment analysis, demonstrating a forward-looking research trajectory. With a cumulative citation count exceeding 390, Shen's work continues to shape the intersection of computer vision, robotics, and affective computing.

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
   One-shot learning based pattern transition map for action early recognition

   32 citations · 2017
6. 6
   A Large-scale Varying-view RGB-D Action Dataset for Arbitrary-view Human Action Recognition

   18 citations · 2019
7. 7
   Robust-EQA: Robust Learning for Embodied Question Answering With Noisy Labels

   17 citations · 2023
8. 8
   Attention Transfer (ANT) Network for View-invariant Action Recognition

   17 citations · 2019
9. 9
   View-invariant action recognition via Unsupervised AttentioN Transfer (UANT)

   12 citations · 2021
10. 10
    Joint Objective and Subjective Fuzziness Denoising for Multimodal Sentiment Analysis

    8 citations · 2024

Key Collaborators

Contact & Links