Victor Zhu's Email

My research and develop experience is mainly focused on the problem of acquiring, representing and understanding visual properties from the real world. I have worked on depth sensors (Stereo cameras, Time-of-Flight sensors, Kinect) for acquiring raw data. I have worked on image statistics, such as sparse, joint and asymmetric filters, to model the image data. I have worked on Hidden Markov Model, Markov Random Fields to analyze the data. I believe a complete system, from acquiring raw data, developing data models and then analyzing the data more effective and efficient, will help scientists and engineers during their course of perceptions, understandings and visualizations. I am still working on these problems and pursue my goal to develop useful technology to assist scientific analysis. Eventually, my ambition is to use them in our daily-life as a product, where the technology becomes a part of our life. I possess many years of programming experience in c/c++ programming and matlab.

Senior Applied Scientist @ Make great things happen for the community and our daily life. From July 2015 to Present (6 months) Senior Computer Scientist @ Research and provide solutions to important data handling, machine learning and compute vision problems. From March 2012 to August 2015 (3 years 6 months) Computer Vision R&D Scientist @ Working on 3D visibility, terrain classification, 3D object recognition problems. From February 2011 to March 2012 (1 year 2 months) Greater Los Angeles AreaVisiting Scholar @ Image Gloss Editing: He worked on the problem of editing image glossy appearance using robust filter. By adjusting the asymmetric statistics of the histogram, the system is able to change the gloss appearance using a single image. Collaborating with Prof. Edward Adelson from MIT and researchers from Microsoft New England (Dr. Ce Liu), Disney Research Lab (Dr. Lavanya Sharan) and University of Central Florida (Dr. Marshall Tappen). Shape from Shading: SFS is a severely ill-posed problem because of the ambiguity looking at a local patch cropped from an image. Human vision will be likely fooled, how can a computer system accomplish this task? By approaching this problem using data driven method, we are envisioned to make step-by-step progress. From August 2010 to January 2011 (6 months) Postdoc Associate @ Shadow system from natural scene: Developed a shadow recognition and removal system from single image captured from natural scenes. The system is successfully trained using >30 processors using MPI. The system and the code src is public available, which has been tested by researchers from CMU and UIUC. Face Recognition: Being a member of a team in automatic tagging faces from facebook images, I am proud to test my algorithm in feature selection (from >80k to 100) using sparsity technique. Although the submission is rejected, I am still working on this problem in competing with other algorithms. Jiejie has also been involved in several other projects, such as activity recognition, saturation removal and image super-resolution thanks to Dr. Marshall Tappen. From January 2009 to July 2010 (1 year 7 months) Postdoc @ Sensor Fusion: Jiejie has focused on the problem of estimating depth from stereo vision and a special type of depth sensor: Time of Flight sensors. By successfully merging them modeled in a Markov Random Field, I have achieved improved depth estimates. This algorithm has been tested in video matting and view synthesis systems, which shows good performance. From September 2007 to December 2008 (1 year 4 months) student @ From 2004 to 2007 (3 years) Project Lead @ 3D Face Modeling on mobile phones: With a team, Jiejie worked on the problem of capturing and modeling a 3D face model using mobile phones from one frontal view. The projects includes two artists from Spain, two network programmers from Singapore and two other graphic engineers from Zhejiang University. I was the team lead for this project. Thanks to Prof. Adrain David Cheok's support, I embraced so many interesting idead. And this also helped me to form the key idea of my dissertation about Multiple View Geometry. From November 2005 to March 2006 (5 months)

PhD, Computer Vision and Human Computer Interaction @ Zhejiang University From 2004 to 2007 Victor Zhu is skilled in: Matlab, C++, Computer Vision, OpenGL, OpenCV, Pattern Recognition, Algorithms, C, Image Processing, Programming, Machine Learning, Data Mining, Optimization, Artificial Intelligence, Computer Science, Python, LaTeX