About Me
Hello, I am Han-Hung Lee and I received my B.S. degree in the Undergraduate Program of Electrical Engineering and Computer Science from NTHU. Currently I am a PhD student at Simon Fraser University under the supervision of professor Angel Xuan Chang. A few of my research interests include 3D Computer Vision, Natural Language Processing and Reinforcement Learning.
Projects
Understanding Pure CLIP Guidance for Voxel Grid NeRF Models
Han-Hung Lee and Angel X. Chang
Project Page- We explore how image augmentations and explicit/implicit voxel grid NeRF models effect pure CLIP guidance.
Graph Neural Networks for 3D Referring Instance Segmentation (IIS Sinica)
Pin-Hao Huang*, Han-Hung Lee*, Hwann-Tzong Chen, Tyng-Luh Liu
Project Page- We use Text-guided Graph Neural Networks with a 3D instance segmentation model to improve the referring instance segmentation task.
- Accepted to AAAI 2021
- We use frequency changes in songs to traverse the latent space of GANs, forming video clips that react to audio to create audio reactive visuals for artistic purposes.
- Accepted to NeurIPS 2019 Workshop on Machine Learning for Creativity and Design
Education
Simon Fraser University
PhD Program in Computing Science
Sept. 2021 - Now
- Supervisor: Angel Xuan Chang (https://angelxuanchang.github.io/)
National Tsing Hua University
B.S. in Undergraduate Program of Electrical Engineering and Computer Science
Sept. 2016 - June 2020
- Overall GPA: 3.87/4.3
- Coursework: Deep Learning, Statistical Learning, Computer Vision, Graphics Programming Game Programming
Experience
Institute of Information Science, Academia Sinica
Undergraduate Research Assistant
July 2019 - July 2021
- Implemented baseline models for neural architecture search methods
- Worked on novel method for 3D referring instance segmentation achieving SOTA results on ScanRefer and ReferIt3D