Yang Hong-Bin (Petingo)
Portfolio
Github
[email protected]
Lausanne, Switzerland
Education
École Polytechnique Fédérale de Lausanne (EPFL)
M.S. in Computer Science
Sep. 2021 ~ Present
- GPA 5.3/6.0
Taiwan Tech
B.S. in Computer Science
Sep. 2016 ~ Jun. 2020
- GPA 4.1/4.3
Czech Technical University in Prague
Exchange Student in the Faculty of Electrical Engineering
Feb. 2020 ~ Jun. 2020
- GPA 1.0/1.0
Skill
Technical - Computer Vision, Machine Learning, Computational Design, Digital Fabrication
Programming - C++, Python, C#, JavaScript, Rhino & Grasshopper
Language - English (C1-C2), German (Beginner), Japanese (Basic), Mandarin (Native)
Professional Experience
IBOIS (Laboratory for Timber Constructions), EPFL
Part-time Student Assistant & Full-time Summer Internship
Apr. 2022 ~ Present
- Contributed to a research project focused on developing augmented reality applications for hand-held carpentry tools.
- Optimize an open-source SLAM method (UcoSLAM) for our task, including:
- Added new features such as map fusion and 3D model reconstruction.
- Ensured compatibility with OpenGL for seamless integration of the virtual scene and camera view.
- Diagnosed and resolved multiple bugs in the original code to enhance its robustness and stability.
Artificial Intelligence and Image Understanding Lab, Academia Sinica
Research Assistant
Jul. 2020 ~ Dec. 2020
- Investigated the use of image quality assessment methods, specifically NIMA and BRISQUE, to improve the realism of composite images created by deep learning-based image harmonization techniques like Deepfake.
- Re-implemented a deep learning-based image forgery detection method (Face X-ray) in PyTorch and experimented with the introduction of various features to enhance its performance.
Lab of Image/Video Processing and Compression, Taiwan Tech
Student Research Assistant
Jan. 2019 ~ Feb. 2020
- Conducted research on chroma subsampling and upsampling of normal and light-field images.
- Proposed novel YUV420 subsampling method with low distortion, achieving a 1.1dB (2.5%) increase in PSNR evaluation compared to the previous state-of-the-art.
STAR BIT Innovation Co.,Ltd.
Software Engineer Intern
Mar. 2018 ~ Jul. 2019
- Created front-end APIs for a decentralized cryptocurrency exchange using JavaScript, enabling five distinct types of Ethereum wallets to interact with the Ethereum blockchain directly through a web interface.
- Partnered with KeyXentic to design and implement a Chrome Extension that extends web3.js support to their hardware cryptocurrency wallet, providing secure and seamless integration with the Ethereum blockchain.
Publication
"Architectural Sketch to 3D Model: An Experiment on Simple-Form Houses," H. B. Yang, M. Johanes, F. C. Kim, M. Bernhard, & J. Huang, CAAD Future 2023
- Utilized Rhino Grasshopper to generate a synthetic dataset of 3D house models and corresponding hand-drawn sketches.
- Trained a deep learning network capable of generating a 3D mesh from a hand-drawn sketch and developed a pipeline to optimize the generated output.
- Constructed a web-based interface for demonstration purposes, allowing users to sketch and view the generated result directly in their browser and export it to Rhino through Grasshopper.
"Effective Chroma Subsampling and Luma Modification for RGB Full-Color Images Using the Multiple Linear Regression Technique," Chung, K. L., Cheng, J. S., & Yang, H. B. , IEEE Access, 8, 118315-118323., Jun 2020 (Link to Paper)
- Proposed the initial idea of applying multiple linear regression on YUV420 subsampling to minimize the distortion.
- Successfully achieved a 1.1dB (2.5%) increase in PSNR evaluation compared to the previous state-of-the-art.
Honors & Awards
2021 | Special Award of the Year – Shin Kong Mitsukoshi International Photography Contest
2020 | Study Abroad Scholarship – Taiwan Tech and Ministry of Education, Taiwan
2019 | College Student Research Scholarship – Ministry of Science and Technology, Taiwan
- Awarded based on the research proposal and academic performance
- Proposed a CNN-based chroma upsampling method for light field images with low distortion, surpassing the conventional bicubic and bilinear methods by 1.24dB (2.9%) and 2.22dB (5.4%), respectively, in PSNR evaluation.
2019 | Silver Award – ICPC Taiwan National Contest for Technology Universities
- Competitive programming contest; preliminary for ACM-ICPC
2018 | 1st Place – National Taiwan University System App Innovation Contest
- Demonstrated a map-based student forum (Android APP, Java)
2018 | Honorable mention – National Collegiate Programming Contest (Taiwan)
- Competitive programming contest; preliminary for ACM-ICPC
2017 | Best Presentation Award – Intelligent Living Hackathon
- Top 6 among 50 teams
- Proposed a management system for a shared refrigerator in a student dormitory using an Android phone and student ID authentication.