Shuyu Wang's personal website

Image-based Relighting Using Implicit Neural Representation.

• Researched on image-based relighting using neural implicit representation and published the thesis on TAMU school server.
• Making use of the sinusoidal representation networks (SIREN) which has periodic activation functions.
• Tested the newly developed algorithm under multiple new lighting conditions. Compared and evaluated the results.

[ The image here depicts the scenes generated by our method under a series of new light positions located between two input images light positions, which creates the look of a single light source slowly transitioning from one position to another. ]

Published Thesis

2D Finite Element Method.

• Implemented 2D triangle finite element method(FEM) on an square object using C++ and OpenGL.
• Tested the movement of the 2D FEM object with both two fixed points and collision with a sphere. Adjusted the stiffness and Poisson’s ratio so that the 2D FEM object shows different physical characteristic when moving.

[ The image here depicts the collision between the created 2D Finite Element Method object and a circle. ]

Demo Page

ECG Signal Data Processing.

• Recorded the Electrocardiograph (ECG) signal data collected through the electrodes using the Bard Electrophysiology Recording System.
• Restored, analyzed, and processed the signal using MATLAB.
• Used low-pass filter to denoise the signal.

[The image here depicts the contrast between the filtered ECG signal (orange) and the original (blue). ]

My Technical Report
About Low-pass filter

Automatic Image Alignment .

• Designed a Python algorithm for feature detecting using the approach of Harris corner detector.
• Designed an algorithm for Random Sample Consensus (RANSAC), where the outliers are iteratively detected and excluded.
• Combined the group work and generated the final algorithm for aligning images based on features detected.
[ The image here depicts the corner features detected from the two input images. ]

Finding Major Tom.

• Project leader. The project generates data and information from 3 APIs: OpenNotify, Gyazo, and Sunrise/Sunset, and calculates the best time for photo shooting the International Space Station.
• Managed the project with a 3-sprints Agile process. Hosted and recorded scrum meetings. Made and managed backlogs, burndown charts, LiveMache designs, meeting agendas and minutes, retrospectives, and presentations. Took charge of daily communicating. Planned and led user study. Participated in designing user interface.

[The image here depicts our search page, which allows user to put in their location and time zone, in order to output the best photoshooting time. ]

Live App on host

HACK computer.

• Constructed a fully functioning general purpose HACK computer using Nand2Tetris logic gates from the ground up.
• The completed computer is able to process HACK assembly language.
• Wrote programs in HACK language.

[The image here depicts the basic logic of the HACK computer. ]

About Hack language
About Nand2Tetris