I-Sheng Fang / Ethan Fang / 方宜晟 / Gî-Tshiânn Png

(“I-Sheng” pronounce like “Ethan”)

[github] [facebook] [instagram] [linkedin] [curriculum vitae] [google scholar]

I am a Research Assistant in CITI at Academia Sinica, working with Dr. Jun-Cheng Chen. Before that, I received my Master degree in Robotics from National Yang Ming Chiao Tung University (the merger of National Chiao Tung University and National Yang Ming University) in January, 2023. My advisors are Prof. Yong-Sheng Chen and Prof. Wei-Chen (Walon) Chiu. I was a graduate student of Department of Computer Science at National Cheng Chi University, Taiwan, working with Prof. Yan-Tsung Peng . I was a research assistants at Enriched Vision Applications Lab, National Chiao Tung University from September 2018 to September 2019, working with Prof. Wei-Chen (Walon) Chiu. I received my Bachelor degree in Mathematical Science from National Chengchi University in January, 2018.

My research interests are in the area of generative model, self/wakly-supervised learning, depth estimation, style transfer, computer vision, and deep learning. I am also interested in their creative application, such as East Asian Ideograph font design, deepfake for education. I believe AI is a leverage, augmenting human ability not only in monotonous tasks but also in content creation.

My personal interests are Typography, Film Photography (check out my instagram @ishengfang), Baseball, Strength and Conditioning Training.

Publications

Camera Settings as Tokens: Modeling Photography on Latent Diffusion Models

I-Sheng Fang, Yue-Hua Han, Jun-Cheng Chen
SIGGRAPH Asia 2024 (Conference Track).
[project page] [demo] [github(comming soon)]

Best of Both Sides: Integration of Absolute and Relative Depth Sensing Modalities Based on iToF and RGB Cameras

I-Sheng Fang, Wei-Chen Chiu, Yong-Sheng Chen
International Conference on Pattern Recognition (ICPR), 2024.
[paper] [supplementary materials] [github]

ES³Net: Accurate and Efficient Edge-Based Self-Supervised Stereo Matching Network

I-Sheng Fang, Hsiao-Chieh Wen, Chia-Lun Hsu, Po-Chung Jen, Ping-Yang Chen, Yong-Sheng Chen
IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2023.
- The 19th Embedded Vision Workshop
  - Best Paper Award
[github][CVPRW2023]

Single Image Reflection Removal based on Knowledge-distilling Content Disentanglement

Yan-Tsung Peng, Kai-Han Cheng, I-Sheng Fang, Wen-Yi Peng, Jr-Shian Wu
IEEE Signal Processing Letters(SPL) Feb. 2022
[IEEE Xplore][github]

Self-Contained Stylization via Steganography for Reverse and Serial Style Transfer

Hung-Yu Chen†, I-Sheng Fang†, Chia-Ming Cheng, Wei-Chen Chiu († indicates equal contribution.)
IEEE Winter Conference on Applications of Computer Vision (WACV), 2020.
[arxiv][github][project page][WACV2020][poster]

Projects

A Century of Heartfelt Sentiment: 100th Anniversary Special Exhibition of the Taiwan Cultural Association

2021/05/08 - 2022/05/08
Exhibition Room D, National Museum of Taiwan Literature, Tainan, Taiwan
Synthesizing DeepFake video for historical figures of Taiwan Cultural Association.
[exhibition page(English)][exhibition page(Mandarin)][exhibition page(Japanese)][press release (English)][press release (Mandarin)][fanpage post]

Education

Master of Science
- The Master’s Degree Program in Robotics, National Yang Ming Chiao Tung University, Taiwan.
  - Merged by National Chiao Tung University and National Yang Ming University on 1, February, 2021.
- September, 2020 - February, 2023
Graduate Student
- Computer Sciences, National Chengchi University.
- January, 2020 - June, 2020
Bachelor of Science
- Mathematical Sciences, National Cheng Chi University, Taiwan.
- September, 2013 - January, 2018

Experience

Research Assistant
- Research Center for Information Technology Innovation(CITI), Academia Sinica, Taiwan.
- March, 2024 -
- Working on generative model with Dr. Jun-Cheng Chen.
Software Engineer Intern
- Microsoft AI R&D Center, Taiwan
- March, 2022 - November, 2022
- Working on vision transformer(ViT), percpetual loss, and generative model with SunDa Yang, Chien-Yi Wang, Prof. Shang-Hong Lai, and Dr. Trista Chen, and the face science team.
Research Assistant
- Enriched Vision Applications Lab, National Chiao Tung University, Taiwan.
- September, 2018 - September, 2019
- Working on style transfer and generative model with Prof.Wei-Chen (Walon) Chiu
Teaching Assistant
- The First Step of AI and Deep Learning with Python, the MOOCs of National Chengchi University.
- July, 2018 - September, 2018
- Teacher is Prof. Yen-lung Tsai.
Fire Fighting Substitute Military Services
- YungPing Branch, Yonghe Squadron, Seventh Crops, Fire Department, New Taipei City Government
- March, 2018 - June, 2018
Deep Learning Scientist
- InQtech Co.,Ltd
- September, 2017 - June, 201
Teaching Assistant
- The First Step of Data Analysis with Python, the MOOCs of National Chengchi University.
- September, 2017 - January, 2018
- Teacher is Prof. Yen-lung Tsai.
Intern
- Leopard Mobile
- July, 2017 - September, 2017
GAN Discussion Host
- NCCU Deep Learning & TWSIAM Student Group
- Winning first place of study group final presentation in NCCU.

Open Source Projects

Relativistic average GAN with Keras, 2018
Spectral Normalization for Keras, 2018
ResNet-DenseNet-Notebook, 2018

Reseach Collections

Typography Research Collection
- Typography is the cross between technology and liberal arts. This page is a research collection that includes computer graphics, computer vision, machine learning that related to typography.
- If anyone found new research, please email it to me :)

Other Research Projects

Font Design with Progressive invariant GANs
- Honorable Mention of MOST Workshop on Generative Adversarial Networks and GAN Project Competition, 2017.
- Introduced GLCM, a invariant, in progressive growing VAE-GAN to embed font in latent space. Generated new font or glyph by modtified the latent space Improved training speed, convergence speed and higher resolution.
Hanzi Typeface Research with Conditional Generative Adversarial Network
- Poster of Taiwan Society for Industrial and Applied Mathematics Annual Meeting, 2017
- Using autoencoder and conditional GAN to embed font in latent space.
- Advisor: Prof. Yen-Lung Tsai.