YE, Ning

Software Engineer

Google Inc.

Mountain View, CA

Listen to my heart, jump into the world.
I am currently a software engineer at Google. Previously, I was a postdoc at Bioinformatics Institute (BII), working on action detection and object detection problems. In March 2011, I graduated as a Ph.D in computer science from National University of Singapore (NUS). During my Ph.D study, I worked with Prof. Terence Sim on face recognition problems. I obtained my B.Sc. degree in computer science from Fudan. My academic training was in computer vision, pattern recognition, image processing and machine learning.
Previously at BII
A Voting Strategy for Sneeze/Cough Detection in Videos
Ning Ye and Li Cheng

We proposed a voting strategy for detecting sneeze and cough in videos. Compared with conventional sliding-window (or its variant family) methods, this idea has two main novelties. First, instances of actions are no longer bounded by 3-D boxes, which allows detection of subjects in locomotion. Second, action detection and tracking are done together. We showed preliminary experimental results of this idea on a surveillance video database.

A Sneeze-Cough Video Database for Vision-Based Flu Surveillance
Ning Ye, Li Cheng, Li Wang, Tuan Hue Thi, Sebastian Maurer-Stroh and Jian Zhang

We proposed the idea of collecting the statistics of flu symptoms in public areas from urban surveillance camera infrastructure by running vision-based action recognition, i.e. sneeze and cough recognition algorithms, on video streams. As the very first step towards this goal, a sneeze-cough video database is collected and baseline experiments are conducted.

Cell Segmentation in Microscopic Images
Li Cheng, Ning Ye, Weimiao Yu and Andre Cheah
We proposed a spoke-like shape feature for detecting and segmenting cells in microscopic images.  By using a combination of this spoke feature, a color feature and a context feature, we are able to segment common microscopic images of stained cells as well as cell images from electron microscopes.
Published in MICCAI 2011 (paper | bibtex)
Previously at NUS
Video Stylization by Single Image Example
Ning Ye, Terence Sim and Xiaoping Miao
We proposed an algorithm to transfer the artistic style of an image, e.g. an oil painting, onto a video.  The algorithm re-synthesizes the texture of the video to mimic the style of the example image, while preserving both spatial and temporal coherence from the original video.
Published in ICIP 2010 (paper 13MB | poster 16MB | bibtex)
Dynamic Facial Signature
Ning Ye, Terence Sim, Li Zhang and Elisa Martínez Marroquín
We proposed several algorithms to extract identity-specific facial motion patterns.  The patterns are strong enough to be used as soft biometric traits.  Once combined with static facial appearance patterns, they provide a feasible way to distinguish identical twins.

Published in CVPR 2010 (paper | poster 7MB | bibtex) and more
Patent: WO 2011/155902

Interactive Portrait Art
Ning Ye and Terence Sim
We developed a real-time facial motion retargeting system.  The system clones the facial expressions of a user to an animated avatar.  The face of the avatar can be automatically extracted from any portrait images or paintings, e.g. Mona Lisa.
Published in WACV 2008 (paper 8MB | poster | bibtex)
Smile :-), Wear It Like a Costume!
Momoyo Torimitsu, Helga Mazyar, Sujun Leow, Ning Ye and Terence Sim
This was an art installation built by Momoyo Torimitsu for ISEA 2008 at National Museum of Singapore.  We advised and provided technical solutions to help make the artistic design come to life.
Tutoring at NUS
AY2009/10 Sem 2
  CS2102/2102s Database Systems
AY2009/10 Sem 1
  CS1102 Data Structures and Algorithm
  GEK1511 Introduction to Computing
AY2008/09 Sem 1
  CS2106 Operating System