Xucong Zhang

Xucong Zhang

Assistant Professor

Research Interest:
My current research is about computer vision, machine learning, and human-computer interactions. My core research interest is human-centred computing as developing techniques for sensing, understanding, and serving the human user.

  • Gaze estimation
    The task is the estimation of where the people looking from the input eye/face image. The methodology is mainly about computer vision and machine learning techniques.

  • Human-computer interaction
    My current research on HCI is applying the computer vision methods, such as gaze estimation, to the HCI task.

  • Human digitalization
    My next five years (2022-2027) research plan is to develop the talking virtual avatar that can act like a human being. The virtual avatar can be used for HCI application and human-related modelling.

  • Biomechanical prior model
    My next ten years (2022 - 2032) research plan is to develop computer vision and machine learning tools for biology- and healthcare-related research. As the starting point, I will introduce the biomechanical prior for the human-related models.

Short Bio:
I am an assistant professor at TU Delft since August 2021. I was a postdoc researcher (2018-2021) in the Advanced Interaction Technologies Lab at ETH Zurich, led by Prof. Otmar Hilliges. I did my PhD research (2013-2018) at Max Planck Institute for Informatics (MPII) (summa cum laude) under the supervision of Prof. Andreas Bulling in Saarbruken, Germany.


  • Computer Vision
  • Human Computer Interaction
  • Machine Learning


  • Doctor of Engineering (summa cum laude), 2018

    Max Planck Institute for Informatics


Towards Single Camera Kinematics Human 3D-Kinematics
Eye Gaze Estimation and Its Applications
Self-Supervised 3D Hand Pose Estimation from monocular RGB via Contrastive Learning
Self-Learning Transformations for Improving Gaze and Head Redirection
Towards End-to-end Video-based Eye-tracking
ETH-XGaze: A Large Scale Dataset for Gaze Estimation under Extreme Head Pose and Gaze Variation
Learning-based Region Selection for End-to-End Gaze Estimation
Gaze Estimation by Exploring Two-Eye Asymmetry
Content-Consistent Generation of Realistic Eyes with Style
Everyday Eye Contact Detection Using Unsupervised Gaze Target Discovery
MPIIGaze: Real-World Dataset and Deep Appearance-Based Gaze Estimation
AggreGaze: Collective Estimation of Audience Attention on Public Displays
Rendering of eyes for eye-shape registration and gaze estimation
Appearance-based Gaze Estimation in the Wild
Robust multi-resolution pedestrian detection in traffic scenes