Dingxi Zhang

Hi!! I'm Dingxi Zhang (张丁夕), an incoming CLS PhD student jointly advised by Prof. Marc Pollefeys (ETH Zurich, CVG) and Prof. Bernt Schiele (Max Planck Institute for Informatics). I obtained my Master's in Computer Science at ETH Zurich, supported by ESOP scholarship. Previously, I completed my undergraduate studies at the University of Chinese Academy of Sciences, where I worked with Prof. Lin Gao and Prof. Shiguang Shan at the Institute of Computing Technology, Chinese Academy of Sciences.

If you are interested in what I do, please reach out. Always happy to chat.

Email  /  CV  /  Github  /  Google Scholar  /  LinkedIn  /  X

Research

My research interests are 3D vision, graphics and robotics. Specifically, my current focus centers on developing efficient neural representations and devising user-friendly and effective methods for interacting with 3D objects and scenes.

A dual-view egocentric benchmark for collaborative human activity understanding, capturing synchronized head-mounted video from two participants along with dense 3D scene reconstructions and high-resolution object scans across three social-reasoning tasks.

A simple yet powerful framework for optical flow and point tracking that significantly improves large displacements estimation and zero-shot generalization.

A 3D neural style transfer framework with adaptable control over perceptual factors based on 3D Gaussian Splatting representation.

A massive annotated dataset capturing 34 hours of bimanual hand activities, totaling 14k motion clips derived from 183 million frames paired with 84k text annotations.

A novel approach in robot navigation combines LLMs with traditional methods.

A novel stroke-based method for editing facial NeRF appearance. Our method outperforms existing 2D and 3D methods in both editing reality and geometry retention.

Using a weight-encoded Neural Network to approximate a grid of latent codes, while sharing the decoder across the entire category to achieve better reconstruction quality while using less memory per sample to store single geometry.

Education

Working Experience

Academic Service

Conference Reviewer: ICRA 2026, CVPR 2026, ECCV 2026

Journal Reviewer: TPAMI, TVCG, Computer Graphics Forum

Other Interesting Projects

Besides research, I also enjoy coding various projects to explore how computer science can enhance our daily lives, as well as foster engaging interactions with diverse subjects.

An efficient and complete system for cervical spine prevention on common PC devices. It can detect cervical posture in real-time, provide timely feedback to remind users to pose correctly, and offers an interactive game to guide the user to exercise their cervical spine scientifically.

A highly compatible and robust platform with impressive performance to construct an artificially organized signaling pathway based on synthetic biology. We designed an NLP-based search tool powered by Llama2 and BERT for iGEM standard parts, Ask NOX.

A system capable of producing nutritious, delicious, and highly user-friendly meal replacement foods using lactic acid bacteria modified by synthetic biology technology. We designed a portable IoT machine to make meal replacement, an application for our machine & an entertaining science game and a convenient tool for creating wiki.

A comprehensive caffeine intake management system that provides users with scientific and proper caffeine intake recommendations. We designed an application Caffeine-monitor to achieve this and an online education platform iGEM EduHub for better Synthetic Biology education.

Website Template: Jon Barron's website.