Projects
Example Input
A collection of face images are provided as input to the framework, on which an encoder is trained.
Face Blending
Feature blending can be achieved based on tuning the encoder input once it is trained.
Face Morphing
Morphing can also be achieved by moving from one vector to another.
Framework for Synthetic Face Blending
Using a PyTorch port of Nvidia's StyleGAN, a framework was created to generate
blends of faces during my internship at UsideU.
An encoder can be trained to map one-hot
encoded vectors that represent selected training
faces to a vector (512 dimensional) in the latent
input space of StyleGAN's synthesis network.
Traffic Simulator + Cooperative Lane-Changing Driving Scheme
For my final-year project in university, I programmed a traffic simulator from scratch in MATLAB and designed an autonomous cooperative driving scheme for efficient lane-changing.
Smart City Model
Constructed a scaled-down model of an intelligent transport system in 2 months. Programmed line-following robots in C++, a vehicle positioning camera program in Python, and an IoT system for inter-vehicular communication to navigate routes and avoid collisions with each other.