Wei Zhang

Software Engineer and Researcher
1600 Amphitheatre Parkway, Mountain View, CA.
Email: coolbee AT google DOT com (business); zhangweiosu AT gmail DOT com (personal)


I am currently a Senior Software Engineer and Researcher at GoogleThe objective of my work is inventing and applying state-of-the-art machine learning and computer vision techniques to provide intelligent online shopping experience. The ambitious and profound long-term goal is developing technical-intensive core engine for revolutionary soft goods shopping service. Brain-storming ideas, solid research experience and on-hand R&D skills are indispensable to achieve this goal. 
Before I joined Google, I worked at Like.com for three months. Before that, I was a Postdoctoral Researcher in the Multimedia Interaction and Understanding Lab (MIUL) of Hewlett Packard Labs. Before HP, I was a graduate student and Ph.D. Candidate in School of Electrical Engineering and Computer Science at Oregon State University. My major advisor is Prof. Thomas G. Dietterich.


Computer Vision; Machine learning; Pattern Recognition; Image Processing; Intelligent User Interface Design; Data Mining, and others.


Silhouette Tagging Project in Google Inc -- Software Engineer (2010 - 2011)
The objective of silhouette tagging is to automatically predicts the silhouette attributes (e.g., mini, long-sleeve, v-neck) of apparel products based on their images and texts. This project is an important part of apparel product search launches,
example. It is very challenging due to the large variations in images and the “semantic gap” in mapping between textual terms and machine-perceivable image features. We develop silhouette tagging algorithm that predicts the silhouette attributes of apparel products with high accuracy. As the first initiator of the project, I researched, designed, implemented and tested most of the feature extraction and classification algorithms. I also participated in the implementation of the overall infrastructure. I am the first author of the WACV 2012 paper [1] and the primary inventor of the two patent applications on this project.

MagicClothing Project in Hewlett Packard Labs – Postdoc (2009 – 2010)
Worked in the Multimedia Interaction and Understanding Lab (MIUL) at HP Labs as a postdoctoral researcher. The objective of my project is to explore various contextual information to significantly improve the performance of automated person clustering in consumer photos. MagicClothing prototype has been implemented and tested to be successful on various family photo datasets. Papers on MagicClothing were published in ICIP [3] and ICME [4]. Four related patent applications have been submitted. In the meantime, I also researched on face recognition in uncontrolled imaging conditions such as in family photos. Two related patent applications have been submitted.

Responsive Mirror Project in Palo Alto Research Center – Intern (2007)
Worked in PARC on Responsive Mirror (RM) project. Responsive Mirror uses multimedia devices, computer vision and machine learning techniques to produce interactive clothing comparison and intelligent clothing retrieval. After six month’s work, a working prototype was implemented and installed within PARC. Evaluation of the product shows that the system improves a lot in the users’ confidence in the decision making and improves their shopping experience. Discovery Channel, BNET and other media have shown great interest on this invention and reported with very positive comments. An introduction paper of the RM system was published as a full paper in IUI 2008 [10]; one companion paper was published in Workshop of CHI 2008 [11]. Another two papers about the system was published in ICDSC 2008 [12] and accepted to ACIVS 2008 [13]. An extended paper on RM system was published in ACM Multimedia Systems Journal 2010 [2]. Three patents have been filed on this invention.

Insect Identification Project in Oregon State University – Ph.D. Candidate (2004 – 2009)
From 2004 to 2009, I worked in the “Insect Identification” project [14] led by Prof. Thomas Dietterich as a Ph.D. candidate. This project seeks to develop devices to automatically imaging and classifying insects with general image processing, machine learning and pattern recognition tools. Focused on learning and recognition, we were facing a very difficult problem posed by the significant intra-class variation and subtle inter-class difference of the appearances of the insects in the captured images. In order to solve the problem, my research was focused on more robust feature extraction, image representation and classification methods. After thorough study and extensive experiments, I proposed PCBR detector [15], unsupervised [17] and supervised [8] visual codebook learning algorithms, a visual codebook evaluation framework [9], a very successful non-redundant codebook learning method [5], a discriminative feature selection algorithm [18], and a codebook-free recognition system [6]. These methods achieved classification accuracy superior to human. The proposed features and the learning methods have also shown to be performing well on generic object recognition datasets.


·         Ph.D. (2004-2009)          Oregon State University (OSU), USA.   

            School of Electrical Engineering and Computer Science

      Advisor: Professor Thomas G. Dietterich

      Ph.D. Dissertation: “Image features and learning algorithms for

      biological, generic and social object recognition”.

·         M.S. (2001-2004)           Xi’an Jiaotong University (XJTU), China. 

      School of Electronic & Information Engineering          

      Advisor: Professor Taiyi Zhang

      M.S. Thesis: “Research on face recognition methods and

      their relations with face databases”.     

·         B.E. (1997-2001)           Xi’an Jiaotong University (XJTU), China.  

      School of Electronic & Information Engineering

      B.S. Thesis: “Design of graphic user interface and network    

      communication in linux”.

Professional Activities and Awards

Invited reviewer of reputed journals including Transactions on Pattern Analysis and Machine Intelligence (PAMI), Elsevier Pattern Recognition, Pattern Recognition Letters, IEEE Transactions on Knowledge and Data Engineering, IEEE Transactions on Circuits and Systems for Video Technology, ACM Multimedia Systems Journal; reviewer of influential international conferences including International Conference on Intelligent User Interfaces, ACM Conference on Image and Video Retrieval.

Joint funding proposals with universities: Organized and prepared joint funding proposals to NSF and IARPA programs with Oregon State University and Brazil University of Federal da Paraíba.

1st Place of 1st Class Award (national) in DSP groups in the 4th Motorola Embedded Processor (MCU/DSP) Design Contest, Shanghai, China, 2002. Won the prize out of 200 competing teams national-wide.

Graduate Research Assistantship, Oregon State University, 2004-2009.

Xinxing Enterprise Scholarship at Xi’an Jiaotong University, 2003.

Excellent Student Scholarships at Xi’an Jiaotong University, 1998, 1999, 2001, 2002.


DISCOVERY CHANNEL Report on “Responsive Mirror” system

BNET Report on “Responsive Mirror” system

Video of ICML 2009 presentation Learning Non-Redundant Codebooks for Classifying Complex Objects


Six patent applications submitted at HP Labs. Five as primary inventor, one as joint inventor.

Three filed patents at Palo Alto Research Center (PARC). Two as primary inventor, one as joint inventor.


    [1] Wei Zhang, Emilio Antunez, Salih Gokturk, Baris Sumengen. Apparel Silhouette Attributes Recognition. In Workshop on the Applications of Computer Vision 2012.

    [2] Wei Zhang, Bo Begole, Maurice Chu. Asynchronous Reflections: Theory and Practice in the Design of Multimedia Mirror Systems. In ACM Multimedia Systems Journal, 2010.

    [3] Wei Zhang, Tong Zhang and Daniel Tretter. Clothing-based person clustering in family photos. In Proceedings of IEEE International Conference on Image Processing (ICIP), 2010.

    [4] Wei Zhang, Tong Zhang and Daniel Tretter. Beyond face: Improving person clustering in consumer photos by exploring contextual information. In Proceedings of IEEE International Conference on Multimedia & Expo (ICME), 2010.

    [5] Wei Zhang, Akshat Surve, Xiaoli Fern and Thomas Dietterich. Learning non-redundant codebooks for classifying complex objects. In Proceedings of the 26th International Conference on Machine Learning (ICML), 2009.

    [6] Gonzalo Martinez-Munoz, Wei Zhang, Nadia Payet, Sinisa Todorovic, Natalia Larios, Asako Yamamuro, David Lytle, Andrew Moldenke, Eric Mortensen, Robert Paasch, Linda Shapiro, Thomas G. Dietterich. Dictionary-Free Categorization of Very Similar Objects via Stacked Evidence Trees. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2009.

    [7] Wei Zhang. Image features and learning algorithms for biological, generic and social object recognition. Ph.D. Dissertation, Oregon State University, 2009.

    [8] Wei Zhang, Thomas G. Dietterich. Learning Visual Dictionaries and Decision Lists for Object Recognition. In Proceedings of the 19th International Conference on Pattern Recognition (ICPR), 2008.

    [9] Wei Zhang, Hongli Deng. Understanding Visual Dictionaries via Maximum Mutual Information Curves. In Proceedings of the 19th International Conference on Pattern Recognition (ICPR), 2008.

    [10] Wei Zhang, Takashi Matsumoto, Juan Liu, Maurice Chu, Bo Begole. An Intelligent Fitting Room Using Multi-Camera Perception. In Proceedings of International Conference on Intelligent User Interface (IUI), 2008.

    [11] Bo Begole, Takashi Matsumoto, Wei Zhang, Juan Liu. Responsive mirror: fitting information for fitting rooms. In Proceedings of the Workshop on Ambient Persuasion at CHI, 2008.

    [12] Wei Zhang, Bo Begole, Maurice Chu, Juan Liu, Nick Yee. Real-Time Clothes Comparison Based on Multi-View Vision. In Proceedings of the 2nd ACM/IEEE International Conference on Distributed Smart Cameras (ICDSC), 2008.

    [13] Wei Zhang, Nick Yee, Juan Liu, Maurice Chu, Bo Begole. Augmented Fashion Exploration Based on Clothes Recognition. Accepted to Advanced Concepts for Intelligent Vision Systems 2008 (ACIVS), 2008.

    [14] Eric Mortensen, Enrique Delgado, Hongli Deng, David Lytle, Andrew Moldenke, Robert Paasch, Linda Shapiro, Pengcheng Wu, Wei Zhang, Thomas Dietterich. Pattern Recognition for Ecological Science and Environmental Monitoring: An Initial Report. N. MacLeod and M. O'Neill (Eds.) Algorithmic Approaches to the Identification Problem in Systematics. 2007.

    [15] Hongli Deng, Wei Zhang, Thomas G. Dietterich and Eric N. Mortensen. Principal Curvature-Based Region Detector for Object Recognition. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2007. (Oral paper, acceptance rate < 10%).

    [16] Congde Lu, Chunmei Zhang, Taiyi Zhang, and Wei Zhang. Kernel Based Symmetrical Principal Component Analysis for Face Classification. Neurocomputing. 2007.

    [17] Natalia Larios, Hongli Deng, Wei Zhang et al. Automated Insect Identification through Concatenated Histograms of Local Appearance Features. Machine Vision and Application (MVA), 2006.

    [18] Wei Zhang, Hongli Deng, Thomas G. Dietterich, Eric N. Mortensen. A Hierarchical Object Recognition System Based on Multi-scale Principal Curvature Regions. In Proceedings of the 19th International Conference on Pattern Recognition (ICPR), 2006.

    [19] Congde Lu, Taiyi Zhang, Wei Zhang and Guang Yang. An Experimental Evaluation of Linear and Kernel-based Classifiers for Face Recognition. Lecture Notes in Computer Science, 2005.

    [20] Congde Lu, Taiyi Zhang and Wei Zhang. Support Vector Domain Classifier Based on Multiplicative Updates. IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences. 2004.


    My LinkedIn profile