Render results of the Synthetic-NeRF (single object) and Replica (indoor scene) datasets.
NeRF rendering engine
Neural renderer built from scratch, built on PyTorch and CUDA C++ extensions.
Hi, I'm Pang.
I am a computer science master student in HKUST working under Dr. Sai-Kit Yeung. I enjoy working with computer vision, graphics and creating fun stuff in general. Some of my recent interests include neural rendering and generative models.
Feel free to check out my latest CV here.
Stuff I worked on
Training NeRF models with loss functions derived from style transfer literature.
Stylization of NeRF scenes
Change the appearance of trained NeRF scenes by applying arbitrary visual styles onto them.
By passing an empty room and locations to insert objects, a decorated scene can be generated.
Neural scene decoration
Conditional GAN designed to transform an empty indoor room into a fully decorated one. Accepted to ECCV 2022.
Demo video of our team's car overtaking other cars.
Autonomous racing
Applied various CNN models to navigate the RC car, given a stream of camera frames. Our team was awarded 2nd place in the 2021 HKUST-Kaisa Autonomous RC-Car Racing Competition.
Our performance in the 2017 Robocon HK contest. (Credits: HKUST Robotics Team)
HKUST Robotics Team
Worked as software engineer for ABU Robocon 2017 Tokyo, mostly on robot positioning control.
Video demonstration of the scanner. (Turn on captions!)
Custom made 3D scanner
A hacky but functional DIY 3D scanner. Link