About me
Hello! I'm a Data Science Master student at ETH Zurich, where I have worked on egocentric vision research projects with Dr. Xi Wang , Prof. Marc Pollefeys at the Computer Vision & Geometry Group, and Prof. Luc Van Gool at the Computer Vision Lab, as well as agriculture-vision projects with Swiss Data Science Center. Before that, I obtained a dual B.S. degree in Data Science and Applied Mathematics from the University of Rochester — Go Yellowjackets! 🐝
My research interests lie in 3D vision and video understanding.
Publications
 | EgoGaussian: Dynamic Scene Understanding from Egocentric Video with 3D Gaussian Splatting Daiwei Zhang, Gengyan Li, Jiajie Li, Mickaël Bressieux, Otmar Hilliges, Marc Pollefeys, Luc Van Gool, Xi Wang Accepted at International Conference on 3D Vision (3DV) 2025 Project / arXiv / Code Recontruct 3D scenes and dynamically tracking 3D object motion from RGB egocentric input alone. |
 | The COVID-19 pandemic and mental health concerns on Twitter in the U.S. Senqi Zhang, Li Sun, Daiwei Zhang, Pin Li, Yue Liu, Ajay Anand, Zidian Xie, Dongmei Li Health Data Science 2022 Paper |
Projects
 | 3D Understanding of Wheat Fields for Phenotypic Trait Extraction Semester project with Swiss Data Science Center (SDSC) and Institute of Agricultural Sciences arXiv in preparation A multi-scale reconstruction pipeline capable of modeling both the entire wheat field and individual wheat heads, using high-throughput RGB data collected with the Field Phenotyping Platform (FIP). |
 | MapSAM: Adapting Segment Anything Model for Automated Feature Detection in Historical Maps Xue Xia, Daiwei Zhang, Wenxuan Song, Wei Huang, Lorenz Hurni Submitted to IEEE Transactions on Geoscience and Remote Sensing Final Project for Research Topics in Cartography arXiv A parameter-efficient fine-tuning strategy that adapts SAM into a prompt-free and versatile solution for various downstream historical map segmentation tasks. |
 | MoE Technique on LLM for Math Reasoning Problems Final Project for Computational Semantics for Natural Language Processing Report Improved the performance of GPT2 and DialoGPT on multi-step mathematical reasoning problems by applying Mixture-of-Experts technique to a novel reasoning scheme called Socratic chain-of-thoughts (CoT). |
 | DeepQSort Data Science Lab project with Empa Code A ViT-based pipeline for efficient wood quality assessment. |