Tian Xu

Research Associate
Computer Laboratory 
University of Cambridge

(Prior to this, I was a postdoc in Institute of Neuroscience & Psychology, 

University of Glasgow. Before that, I obtained my PhD from School of 

Computing Science, University of Glasgow.)


Email:  tian.xu AT cl.cam.ac.uk


Research  Interests
Computer Vision      Developing algorithms for facial expression and action unit 
                                         recognition from image sequences.
                                       Researching on how to optimize image processing algorithms 
                                         such as stereo matching for robotic vision. 
Deep Learning:          Exploiting psychophysical methods to understand mechanisms of 
                                         deep neural network for face recognition.
Parallel Computing: Investigating how to accelerate algorithms for real-time 
                                       applications by utilizing multi-core CPU and GPU. 

Tian Xu, Jiayu Zhan, Oliver G.B. Garrod, Philip H.S. Torr, Song-Chun Zhu, Robin A.A. Ince, Philippe G. Schyns, Deeper Interpretability of Deep NetworksarXiv preprint arXiv:1811.07807 (2018).

Tian Xu, Oliver G.B. Garrod, Steven H Scholte, Robin A.A. Ince, Philippe G Schyns, Using Psychophysical Methods to Understand Mechanisms of Face Identification in a Deep Neural NetworkComputer Vision and Pattern Recognition Workshops (CVPRW), Salt Lake City, US, 2018. [SuppMat]

Tian Xu, Oliver G.B. Garrod, Chaona Chen, Rachael E. Jack, Philippe G. Schyns, Deep Neural Network Identifies Dynamic Facial Action Units from Image Sequences. Vision Sciences Society. 2018.

Oliver G.B. Garrod, Tian Xu, Philippe G. Schyns, Understanding Information Processing Mechanisms for Face Categorizations in Deep Neural Networks. Vision Sciences Society. 2018.

Tian Xu, Oliver G.B. Garrod, Lukas Snoek, Steven H Scholte, Philippe G Schyns, Using Psychophysical Methods to Study Face Identification in a Deep Neural Network. (2017) Journal of Vision. 17. 248. 10.1167/17.10.248.

Tian XuEfficient and accurate stereo matching for cloth manipulationPhD Dissertation, University of Glasgow, 2016.

Tian Xu and W. Paul Cockshott, Evaluation of Foveated Stereo Matching for Robotic Cloth Manipulation. 11th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISAPP 2016), Rome, Italy, 2016.

Tian Xu and W. Paul Cockshott, Guided Filtering Based Pyramidical Stereo Matching for Unrectified Images. International Conference of Image and Vision Computing New Zealand (IVCNZ 2015), Auckland, New Zealand, 2015. (Best Paper Award)

Mozhgan Chimeh, W. Paul Cockshott, Susanne B. Oehler, Ashkan Tousimojarad and Tian Xu. Compiling Vector Pascal to the XeonPhi. Concurrency and Computation: Practice and Experience, pages 5060-5075, 2015.

Tian XuW. Paul Cockshott, and Susanne Oehler. Acceleration of Stereo-Matching on Multi-core CPU and GPUIn: 16th IEEE International Conference on High Performance Computing and Communications (HPCC 2014), Paris, France, 2014.

W. Paul Cockshott, Susanne Oehler and Tian XuDeveloping a compiler for the XeonPhi (TR-2014-341). Technical Report. University of Glasgow, Glasgow, 2014. 

W. Paul Cockshott, Susanne Oehler, Tian Xu, J. Paul Siebert, and Gerardo Aragon-Camarasa. Parallel Stereo Vision AlgorithmMany-core Applications Research Community Symposium (MARC 2012), pages 45-50, Aachen, Germany, 2012.
Tian Xu, and W. Paul Cockshott. Multi-core Efficiency Optimization for Pyramid Algorithm in Image Processing, Poster, International Computer Vision summer school (ICVSS), Sicily, Italy, July 2012