|
Romrawin (Jin) Chumpu I'm a PhD student in Computer Science at NUS Computing and the Singapore International Graduate Award (SINGA) 🦁 with A*STAR, (I'm the luckiest student to be) supervised by Jane E at the palette lab 🎨 and Fayao Liu at A*STAR Institute of Advanced Intelligence and Computing (IAIC) 🤖. My research interests now span physics-based 3D generation and animation. I focus on developing physics-based learning systems for realistic motion and interaction, while designing tools that better support animators. My alter ego is an all-rounded artist 🎨 (professional in fine arts). I love to do art in all forms, from traditional painting/crafting to 3D digital art. |
Non mi legga chi non e matematico.Let no one read me who is not a mathematician. - Leonardo da Vinci |
|
🧮 My Journey from Mathematics
I'm from Chiang Mai, Thailand. When I was 14, I was selected as a Mathematics scholar through the Junior Science Talent Project (JSTP) by NSTDA and the SCiUS scholarship program by the Ministry of Science. I feel really fortunate to have had both educational support and my own research funding throughout my BEng/MSc studies at SIIT. Back then, I was fascinated by applied math in nonlinear and differential equations, so my projects focused on epidemic models, multiphase fluid and materials simulation. ⛄ I also spent time at the University of Waterloo twice - once as a research intern and once as an exchange student in 2018–2019. 🌲 More recently, in the summer of 2023, I was a visiting student at Stanford University, and I absolutely loved the mountains there!
✨ About My Multilingual Name
My Thai name spells รมย์รวินท์ ชุมภู and "ร" or "R" sound is the hardest Thai letter to pronounce! My Thai nickname is จิ้นปู้ (Jinpu) which can pronouce in Chinese to 进步. (In Chinese this word means "progress" or "advancement".) But this is even harder to pronounce so I shorten my name to "Jin". This works for quite a while now😄 |
|
✉️ romrawin[at]u.nus.edu 📄 Google Scholar 🧑🏻💻 Github 📑 Visual Notes 🎨 Art Gallery |
News |
|
Selected PublicationsSee full list of publications on Google Scholar |
|
|
|
When Better Teachers Don't Make Better Students: Revisiting Knowledge Distillation for CLIP Models in VQA
Pume Tuchinda, Parinthapat Pengpun, Romrawin Chumpu, Patomporn Payoungkhamdee, Sarana Nutanong, Peerat Limkonchotiwat 4th International Workshop on Multimodal and Responsible Affective Computing (MRAC), ACM Multimedia, 2026 Paper |
|
Anthropogenic Regional Adaptation in Multimodal Vision-Language Model
Samuel Cahyawijaya, Peerat Limkonchotiwat, Tack Hwa Wong, ⋯, Romrawin Chumpu, et al. (48 authors) Annual Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and International Joint Conference on Natural Language Processing (AACL-IJCNLP), 2026 Paper |
|
Distilling Multilingual Vision-Language Models: When Smaller Models Stay Multilingual
Sukrit Sriratanawilai, Jhayahgrit Thongwat, Romrawin Chumpu, Patomporn Payoungkhamdee, Sarana Nutanong, Peerat Limkonchotiwat arXiv preprint, 2025 Paper |
|
Physics-informed graph neural networks accelerating microneedle simulations towards novelty of micro-nano scale materials discovery
Romrawin Chumpu, Chun-Lin Chu, Tanyakarn Treeratanaphitak, Sanparith Marukatat, Shu-Han Hsu Engineering Application of Artificial Intelligence, 2023 Project | Paper | Code | BibTex |
|
Synthetic face generation from in-the-wild face components swapping Romrawin Chumpu, Pitchayagan Temniranrat, Sanparith Marukatat 17th International Joint Symposium on Artificial Intelligence and Natural Language Processing (iSAI-NLP), 2022 Paper | Code Presented online at OAMLS: Online Asian Machine Learning School as a part of 14th Asian Conference on Machine Learning (ACML 2022), Hyderabad, India. Poster |
|
The association between dengue incidences and provincial-level weather variables in Thailand from 2001 to 2014
Romrawin Chumpu, Nirattaya Khamsemanan, Cholwich Nattee PLOS One, 2019 Paper | BibTex |
Teaching |
|
Modern Image Search
Romrawin Chumpu Super AI Engineer Development Program Season 4, January, 2024 Course Page | Slide | AIAT MOOC An introductory course on modern image search, including introduction of SOTA multimodal models and hands-on coding tutorials. 
|
TV Show |
|
Thailand's Super AI Engineer the Reality Show
One of the ten cast members selected from 8000 competitors nationalwide Representative of woman in AI (3/10) and a role in AI commutiy as a researcher 🌟 Finalist Team On-air via national TV channel in 5/5 episodes from November 2022 to January 2023 Full Episodes | Annoucement | Teaser | News Cast Members | Ad Poster |
|
This page adapted from Jon Barron's site. |