Welcome to Dahua's Web Home

Thoughts sparkle by sharing

Welcome

Welcome to my research page. This page introduces my works in academic research as well as my papers and theses. It is my pleasure to discuss research problems with you on statistical learning, computer vision, and related fields.

As a research student, I sincerely hope that my efforts can contribute to the further advancement of the fields that I am working on, as well as the technological innovation that improves our everyday life.

My Research Interests

Research Fields

Statistical Learning
A family of approaches to machine intelligence based on statistical modeling of data and their relations. They learn statistical models with training data, which can be then applied to new samples for statistical inference. It is one of the most active fields in artificial intelligence.
Computer Vision
Computer vision is concerned with the technologies that can obtain useful information from images and videos. It covers a broad spectrum of topics, from low level sensoring to high level image modeling. In recent years, machine learning techniques are playing an increasingly important role in computer vision.
Pattern Recognition
Pattern recognition is a field investigating how to classify data according to the observed features based on a recognition model capturing the prior information. A typical pattern recognition system consists of three stages: data aquisition, description and classification. The description and classification has a close connection with the statistical learning (supervised learning) theories and techniques.
Vision-based Graphics
Computer graphics is a fascinating field that uses computer to render visual scenarios. Traditionally, the graphics systems are designed by human and difficult to achieve photorealistic quality. Researchers gradually find that computer vision models would offer great help to image and video synthesis. Consequently, vision-based graphics, which integrates the analysis in vision and synthesis in graphics, emerged and soon evolved into an active field.
[Back to Top]

Research Topics

  • Multi-factor Expression Analysis and Synthesis
  • Face Recognition and Verification
  • Face Superresolution and Sketch Synthesis
  • Discriminative Feature Extraction
  • Information Theory Based Learning
  • Integration of Generative Learning and Discriminative Learning
  • Subspace Learning, Kernel Learning, Manifold Embedding and their relations to spectral theory
  • Coupled Learning: build models connecting associative sample spaces
  • Ensemble Learning: deploy and integrate multiple models
  • Context Modeling: model the interactions and relations between different objects in the same environment
[Back to Top]

My Publications

Year 2007

Dahua Lin, and Xiaoou Tang. Quality-Driven Face Occlusion Detection and Recovery. Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2007 (CVPR 2007). [pdf]

Zhifeng Li, Dahua Lin, and Xiaoou Tang. Discriminant Mutual Subspace Learning for Indoor and Outdoor Face Recognition. Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2007 (CVPR 2007).

[Back to Top]

Year 2006

Dahua Lin, Shuicheng Yan, and Xiaoou Tang. Pursuing Informative Projection on Grassmann Manifold. Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2006 (CVPR 2006), Volume 2, Pages:1727-1734.(Accepted for Oral presentation). [pdf]

Dahua Lin, and Xiaoou Tang. Recognize High Resolution Faces: From Macrocosm to Microcosm. Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2006 (CVPR 2006), Volume 2, Pages:1355-1362. [pdf]

Dahua Lin, and Xiaoou Tang. Conditional Infomax Learning: An Integrated Framework for Feature Extraction and Fusion. Proceedings of 9th IEEE European Conference on Computer Vision, 2006 (ECCV 2006), Part I, Pages:68-82.(Accepted for Oral presentation). [pdf]

Dahua Lin, and Xiaoou Tang. Inter-Modality Face Recognition. Proceedings of 9th IEEE European Conference on Computer Vision, 2006 (ECCV 2006), Part I, Pages:13-26.(Accepted for Oral presentation). [pdf]

[Back to Top]

Year 2005

Dahua Lin, and Xiaoou Tang. Coupled Space Learning for Image Style Transformation. Proceedings of 10th IEEE International Conference on Computer Vision, 2005 (ICCV 2005), Volume 2, Pages:1699 - 1706. [pdf]

Wei Liu, Dahua Lin, and Xiaoou Tang. Hallucinating Faces: TensorPatch Super-Resolution and Coupled Residue Compensation. Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2005 (CVPR 2005), Volume 2, Pages:478 - 484.(Accepted for Oral presentation). [pdf]

Zhifeng Li, Wei Liu, Dahua Lin, and Xiaoou Tang. Nonparametric Subspace Analysis for Face Recognition. Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2005 (CVPR 2005), Volume 2, Pages:961 - 966. [pdf]

Dahua Lin, Shuicheng Yan, and Xiaoou Tang. Feedback-based Dynamic Generalized LDA for Face Recognition. Proceedings of IEEE International Conference on Image Processing, 2005 (ICIP 2005), Volume 2, 2005. Pages:922 - 925. [pdf]

Dahua Lin, Shuicheng Yan, and Xiaoou Tang. Comparative Study: Face Recognition on Unspecific Persons using Linear Subspace Methods. Proceedings of IEEE International Conference on Image Processing 2005 (ICIP 2005), Volume 3, Pages:764 - 767. (Accepted for Oral presentation). [pdf]

Dahua Lin, Wei Liu, and Xiaoou Tang. Layered Local Prediction Network with Dynamic Learning for Face Super-resolution. Proceedings of IEEE International Conference on Image Processing, 2005 (ICIP 2005), Volume 1, Pages:885 - 888. [pdf]

Dahua Lin, Yingqing Xu, Xiaoou Tang, and Shuicheng Yan. Tensor-based Factor Decomposition for Relighting. Proceedings of IEEE International Conference on Image Processing, 2005 (ICIP 2005), Volume 2, Pages:386 - 389. [pdf]

Wei Liu, Dahua Lin, and Xiaoou Tang. Face Hallucination Through Dual Associative Learning. Proceedings of IEEE International Conference on Image Processing, 2005 (ICIP 2005), Volume 1, Pages:873 - 876. [pdf]

Wei Liu, Dahua Lin, and Xiaoou Tang. Neighbor Combination and Transformation for Hallucinating Faces. Proceedings of IEEE International Conference on Multimedia and Expo, 2005 (ICME 2005). Pages:145 - 148. (Accepted for Oral presentation). [pdf]

[Back to Top]

My Theses

M.Phil Thesis (CUHK)

Dahua Lin. Discriminant Feature Pursuit: From Statistical Learning to Informative Learning. For the Degree of Master of Philosophy in Information Engineering, The Chinese University of Hong Kong. (Supervised by Prof. Xiaoou Tang) [pdf]
This thesis won the Engineering Faculty's Outstanding Thesis Award, which is awarded to only one M.Phil thesis and one Ph.D thesis in the whole engineering school each year.

Undergraduate Thesis (USTC)

Dahua Lin. Multi-factor Decomposition and Analysis for Facial Expression. For the Degree of Bachelor in Electronic Engineering and Information Science, The University of Science and Technology of China. (Supervised by Dr. Yingqing Xu, and Dr. Houqiang Li)
This thesis was completed in Microsoft Research Asia. It won the University-wide Outstanding Undergraduate Thesis Award.

[Back to Top]

My Research Experience

Face Recognition Systems in Real Applications

The Chinese University of Hong Kong. (2004 - 2007),
Supervised by Prof. Xiaoou Tang.

  • Lead the development team since Mar, 2006
  • Develop the core recognition algorithms for the project
  • Develop the face recognition SDK and its core library
  • Develop a face recognition-based photo album software
  • Develop a real-time motion tracking system
  • Cooperate with USTC on a image data collection project
[Back to Top]

On Statistical Learning, Computer Vision, and Pattern Recognition

The Chinese University of Hong Kong. (2004 - 2007),
Supervised by Prof. Xiaoou Tang.

  • Statistical feature extraction and dimension reduction
  • Information theory-based statistical learning and feature extraction
  • Face super-resolution
  • Model-based image style transformation
  • High-resolution face recognition
  • The connections between different statistical learning models
  • Context modeling and context-relevant album organization
[Back to Top]

On Vision-based Synthesis

Microsoft Research Asia. (2004),
Supervised by Dr. Yingqing Xu.

  • Investigate a variety of face modeling and statistical learning methods
  • Develop softwares for facial expression analysis, modeling, and synthesis
[Back to Top]

On Dynamic 3D Reconstruction and Stereo Tracking

The University of Science and Technology of China. (2003),
Supervised by Dr. Houqiang Li.

  • Develop a novel framework for object tracking under stereo surveillance system based on dynamic 3D position estimation techniques
  • Develop a series of techniques to enhance the tracking efficiency and robustness
[Back to Top]

My Academic Services

Paper Reviewer of Journals

  • IEEE Transaction on Pattern Analysis and Machine Intelligence (PAMI), 2005-2006
  • Electronic Letters on Computer Vision and Image Analysis(ELCVIA), 2006-2007
  • Neural Computation, 2006

Program Commitee Member of Conferences

  • International Conference on Computer Vision (ICCV 2007)
  • IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2007)
  • Asian Conference on Computer Vision (ACCV 2007)

[Back to Top]