Po-Chen Wu (吳柏辰)

Facebook Reality Labs

I am a research scientist at Facebook Reality Labs working on hands & interaction tracking.
I received my Ph.D. degree in the Graduate Institute of Electronics Engineering at National Taiwan University in 2018 advised by Prof. Shao-Yi Chien in the Media IC & System Lab. I was also co-advised by Prof. Ming-Hsuan Yang from University of California, Merced since 2015.
Before being a graduate student, I received my B.S. degree in the Department of Electrical Engineering at National Taiwan University in 2010.

Here is my CV.

News

06/2019 Having the honor to win the Outperforming Alternative Military Serviceman Award.
01/2019 Having the honor to win the NTU GIEE Best Ph.D. Dissertation Award.
08/2018 Having the honor to win the IPPR Best Ph.D. Dissertation Award.
05/2018 I have passed my oral defense!
05/2018 Having the honor to become an Honorary Member of The Phi Tau Phi Scholastic Honor Society.
03/2018 Our paper "Direct Pose Estimation for Planar Objects" is accepted by CVIU.
03/2018 I have accepted the job offer from Facebook Reality Labs in the position of Research Scientist. Before going to work there at Redmond in 2019, I will do the military service in Taiwan for one year.
10/2017 Our paper wins Honorable Mention Award at UIST 2017.
09/2017 During 09/07, I am invited to give a talk on "DodecaPen: Accurate 6DoF Tracking of a Passive Stylus" at Students in AI Workshop, Hsinchu, Taiwan. Slides
08/2017 During 08/09, I am invited to give a talk on "DodecaPen: Accurate 6DoF Tracking of a Passive Stylus" at TAI CHI 2017, Tainan, Taiwan. Slides
07/2017 Our paper "DodecaPen: Accurate 6DoF Tracking of a Passive Stylus" is accepted by UIST 2017.
07/2017 Our paper "A Benchmark Dataset for 6DoF Object Pose Tracking" is accepted by ISMAR-Adjunct 2017.

Research

Dissertation

Accurate 6DoF Object Pose Estimation and Tracking

Po-Chen Wu
Ph.D. Dissertation
Graduate Institute of Electronics Engineering, National Taiwan Univeristy, 2018
*** NTU GIEE Best Ph.D. Dissertation Award ***
*** IPPR Best Ph.D. Dissertation Award ***
Dissertation Slides

Real-time Eye Localization, Blink Detection, and Gaze Estimation System without Infrared Illumination

Bo-Chun Chen, Po-Chen Wu, and Shao-Yi Chien
IEEE International Conference on Image Processing (ICIP), 2015
Paper

Stable Pose Tracking from a Planar Target with an Analytical Motion Model in Real-time Applications

Po-Chen Wu, Yao-Hung Tsai, and Shao-Yi Chien
IEEE International Workshop on Multimedia Signal Processing (MMSP), 2014
Paper Supp

Digital Circuit Lab

I had been the TA of Digital Circuit Lab at NTUEE for two years and I made some slides for the courses. Feel free to download them if they are helpful to you

Topic Keywords PPT PDF
[1]Verilog Coding Guidelines Verilog syntax, Coding style
[2]Verilog Simulation & Debugging Tools NC-Verilog, nLint, nWave, Verdi
[3]My First FPGA for Altera DE2-115 Board Megafunction, Pin assignment, SDC file
[4]My Second FPGA for Altera DE2-115 Board System builder, ModelSim-Altera
[5]DE2-115 Control Panel - Part I LED, 7-segment Display, LCD Display
[6]DE2-115 Control Panel - Part II Memory, VGA, IR, USB, RS232, HSMC
[7]Memory Devices on DE2-115 SRAM
[8]16x2 LCD Module on DE2-115 CFAH1602B-TMC-JP
[9]24-bit Audio CODEC Audio signal, Datasheet
[10]Introduction to VGA Timing specification
[11]My First Nios II for Altera DE2-115 Board Nios II IDE, CFI flash
[12]Introduction to PG & LA Programm data generator, Logic analyzer
[13]2.3 Inch 8x8 Dot-Matrix Display Xiaoluren
[14]Image and Audio File Formats WAV, JPEG, GIF, PNG, BMP
[15]RSA-256bit Cryptography, RSA, Montgomery algorithm

Misc


• Software

KinectV2Recorder

KinectV2Recorder

KinectV2Recorder is a Win32 program used to record the Color, Depth, and Infrared sequences captured from Kinect V2. These three different sequences are somehow synchronized in hardware (30 fps). The depth and infrared frames are captured in the exact same time, followed by the color frame (~ 6 ms later).

ObjViewer

ObjViewer

ObjViewer is a C++ program used for viewing .OBJ 3D model files. An .OBJ file can be accompanied by .MTL texture files. A .MTL file can be accompanied by other image files. Users can also generate image sequences with a 3D model rendered according to specified poses.


• Tutorials

Topic Keywords PPT PDF
Principles of Machine Learning Classification, Regression, Regularization
Matrix Calculus Notation, Derivative formulas, Chain rule
Introduction to AR Camera calibration, Marker detection, Pose estimation
Feature-based Pose Estimation Feature Matching, RANSAC, PnP algorithms
Introduction to Kalman Filter Overview, Theory, Example
KUKA User Manual Programmable robotic arm, KUKA KR 16-2
How to Present (Chinese version)

• Lab Meeting Slides

All slides can be found here. You may need to ask me for the download password.