Wei Pan
AI-driven algorithms and agents that see in 3D and ship in industry
R&D Director / PI at OPT Machine Vision (Japan). Ph.D. from SUTD and B.Eng. from Zhejiang University. I turn geometric perception research into factory-grade industrial intelligence: around 40 publications across CVPR, ECCV, EAAI, Patterns, and Advanced Photonics, plus deployed systems for 3D imaging, inspection, measurement, registration, and robot vision.
Featured Work
All publications →
SuP: Sub-cloud Driven Point Cloud Registration
Generative deep-learning-embedded asynchronous structured light for three-dimensional imaging
An Improved Graph Attention Network for Semantic Segmentation of Industrial Point Clouds in Automotive Battery Sealing Nail Defect Detection
Professional Service
Full CV →About
Full CV →I came to vision through optics — first learning how light is shaped, then how shape can be recovered from light. Vision keeps reminding me that seeing is not understanding: a camera records light, and meaning comes from the questions we bring to it.
Much of my work is computation: algorithms that turn light into geometry, and learning that fills in what physics alone cannot. AI has changed how fast we can try ideas, not what makes them right. A model that fits the data is a hypothesis; an algorithm that survives the production line has earned something more.
Hangzhou, Singapore, Shenzhen, Tokyo: moving between places, and between academia and industry, has made me wary of single answers. What I care about most is the quiet part of the craft — being precise, being patient, and being willing to be wrong. Look carefully, change one thing, check again.