Projects

• Detecting and Recognizing Texts in Images of Display Ads
  Research Intern @ Microsoft Research
  Developed algorithms for text detection and recognition in complex images of display ads.
• Object Detection and Segmentation in Traffic and Retail Surveillance Videos
  Research Intern @ IBM T. J. Watson Research Center
  Working on fraud detection in retail surveillance videos and abandoned object detection in public surveillance videos.
• Joint Detection and Segmentation of Cell Regions in Microscopic Images
  Research Assistant @ Carnegie Mellon University
  Developed a novel heterogeneous CRF model that realizes joint detection and segmentation of cell regions in microscopic images. The proposed model makes it possible to distinguish among individual cells with a CRF by eliminating inherent redundancy while maintaining full expressiveness.
• Adaptive Cell Detection under Phase Contrast Microscopy
  Research Intern @ Intel Research Pittsburgh
  Developed a novel learning-based algorithm that adapts to different types of cells and imaging conditions to achieve robust detection under various scenarios without human intervention.
• Cognitive State Classification using Magnetoencephalography Data
  Ph.D. student @ Carnegie Mellon University
  Developed signal processing and machine learning algorithms that successfully decoded human cognitive states from Magnetoencephalography data, and moreover, revealed some important phenomena of human brain activities.
• Object Recognition by Discreet and Continuous Search
  Research Assistant @ Carnegie Mellon University
  Developed a highly effective part-based object detection algorithm that integrates discreet and continuous search to enhance robustness to noise and occlusion while reducing computational complexity.
• Visual-Tracking Robot
  R&D Intern @ National Instruments China
  Developed a robot which could keep its eye (a video camera) gazing at any designated moving object (by rotating the camera) with high robustness and low latency. This was a first-of-its-kind robotic system integrating vision and motion components of NI's LabVIEW platform and products.
• Robust Visual Object Tracking against Drift and Occlusion
  Research Assistant @ Fudan University
  Developed a set of visual tracking algorithms capable of inhibiting template drift and handling various types of occlusions robustly. This work pushed forward the state of the art in visual object tracking.
• DSP-Based Real-Time Visual Tracking System
  Research Assistant @ Fudan University
  Developed a real-time visual tracking algorithm tailored to a TI DM642 DSP board. This work successfully fulfilled the requirements of a major national project.
• Intelligent Server Power Management
  Software Engineer Intern @ IBM China Development Laboratory
  Developed a novel server power management strategy that enables a server to automatically choose optimal throttling settings over time by adaptively learning the relationship between power consumption and CPU throttling states under various workloads. (IBM Invention Disclosure)
• Multimedia Experiment Platform for Human Psycho-Visual Modeling
  Research Assistant @ Fudan University
  Developed the platform, organized experiments, and gained intriguing insights from the experimental data.
• Adaptive GOF Residual Operation in Video Compression
  Research Assistant @ Fudan University
  Designed a novel 3D-DWT video coding algorithm which significantly enhances compression efficiency by performing content-adaptive GOF residual operations.
• Real-Time Video Compression and Transmission System   (National Project 863)
  Research Assistant @ Fudan University
  Implemented 3D-DWT video compression algorithm on an ADI BF533 DSP and designed an interface between video camera and main system. This work successfully fulfilled the requirements of a major national project.