Real-time 3D Motion Capture: A real-time vision-based system for capturing 3D human motion. See video of our SIGGRAPH98 demonstration. The project was a collaboration project among University of Maryland, ATR (Japanese research center), and Media Lab (MIT) (Students who are interested in working in this research, please contact me.>).
	Vision-based Surveillance System: We analyze surveillance videos to detect and track humans and trying to recognize activities in scene.
	Mobile Translator: Natural scene text translation on mobile.
	Mobile Locator: Localization and navigation using mobile imaging.
	Background Subtraction: Robust methods for segmenting moving foreground objects from the background scene.

PAST AFFILIATED PROJECTS & LABS:

	Vision-based Interactive Games: We built 2D/3D interactive games that uses player motion to control the character. No need active sensor devices, we need just a webcam. A part of games we built are Disc Devil (TICTA award), (the 2nd Runner Up Microsoft Imagin Cup), an Interactive Aerobic Training System, etc.
	We applies image processing techniques in analyzing agricultural products (mangoes, beans, rice) for automatic inspection and sorting. These projects were collaborated with the Department of Agricultural Engineering, KMITL.
	HID:Human Identification at a Distance Project.
	ADVICE: Activity Detection by Video Content Extraction Project. This project is a part of ARDA's VACEProgram.
	PIRL:Perceptual Interfaces and Reality Laboratory.
	Keck Laboratory for the Analysis of Visual Motion: The Keck lab is a multi-perspective imaging laboratory, containing 64 digital, progressive-scan cameras organized as sixteen short baseline stereo rigs. For more information of the lab specifications, click here.
	3D Head Pose Estimation: A method for head tracking and pose estimation.
	Handwritten Recognition: A method of unconstraint handwritten recognition using Dual-Cooperative Neural Networks.

COURSE PROJECTS:

	Virtual Video Teleconferencing: This project was for Advanced Topic in Computer Network course taught by Professor Ashok Agrawala at UMD in 1995. We employed our expertise in image processing and computer vision to design and implement a real-time system for video teleconferencing. This work was a team project with Dr.Ross Cutler (Now at Microsoft Research Lab, USA) , Dr.Ismail Haritaoglu (Now at IBM Research Lab, USA), and Dr.Vasanth Philomin (Now at Phillips Research Lab, Germany).
	Content-based Image Query: This project was for Advanced Topic in Database System course taught by Professor Christos Faloutsos in 1996. We employed image feature extraction, wavelet encoding and R3 tree structure to build a system for image indexing and query. This work was a team project with Dr. Songrit Maneewongvatana (Now at KMUTT), and Dr. Charnyote Pluempitiwiriyawej (Now at Mahidol U.).
	Maximum-Flow Finder Applet: This project was for Combinatorics and Graph Theory course taught by Prof.Duane A. Cooper in 1996. This is a Java applet which allows the user to create a network along with edge capacity and then run the Ford-Fulkerson algorithm. User also can see how the algorithm progresses step-by-step. (Try it! from these links: Professor Kirk Pruhs at UPitt Page).